[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[no subject]
|
From: |
Patrice Dumas |
|
Date: |
Mon, 12 Feb 2024 18:07:19 -0500 (EST) |
branch: master
commit fc1807d84addf5a2ddee2ae8c00306b51efe0b8a
Author: Patrice Dumas <pertusus@free.fr>
AuthorDate: Tue Feb 13 00:05:38 2024 +0100
* tp/Texinfo/options_data.txt (DOCUMENTLANGUAGE_COLLATION)
(XS_STRXFRM_COLLATION_LOCALE): add customization variables.
* tp/Texinfo/XS/main/call_perl_function.c (call_setup_collator)
(call_collator_getSortKey), tp/Texinfo/XS/main/document_types.h
(BYTES_STRING, enum collation_type_name),
tp/Texinfo/XS/main/manipulate_indices.c (INDEX_COLLATOR, get_sort_key)
(setup_collator, compare_byte_strings, compare_index_letter)
(compare_sortable_subentry_keys),
tp/Texinfo/XS/main/manipulate_indices.h (SORTABLE_INDEX_SUBENTRY):
add an enum for the types of collations, and a structure for a
collator. Add the possibility to call perl code to get a collator
object and get Perl sort keys as a possibility for collation sort
strings. Use bytes strings with length through the BYTES_STRING
structure, to be able to use strings with NUL within as collation
string keys. This is needed for collation keys coming from Perl and a
logical structure in C for collation binary strings. Use memcmp to
compare.
* tp/Texinfo/XS/main/document.c (document_indices_sort_strings)
(destroy_document_information_except_tree),
tp/Texinfo/XS/main/document_types.h (INDEX_SUBENTRY_SORT_STRING)
(INDEX_ENTRY_SORT_STRING, INDEX_SORT_STRINGS, INDICES_SORT_STRINGS)
(DOCUMENT), tp/Texinfo/XS/main/manipulate_indices.c
(destroy_index_entries_sort_strings),
tp/Texinfo/XS/main/manipulate_indices.c
(setup_index_entries_sort_strings, setup_sortable_index_entries)
(setup_sort_sortable_strings_collator, sort_indices_by_letter): add
setup_index_entries_sort_strings based on setup_sortable_index_entries
code and add setup_sort_sortable_strings_collator, as in Perl. Add
document_indices_sort_strings.
* tp/Texinfo/XS/main/manipulate_indices.c
(destroy_indices_sortable_entries): add.
* tp/Texinfo/XS/main/document.c (new_collation_sorted_indices)
(find_collation_sorted_indices, sorted_indices_by_letter)
(destroy_document_information_except_tree),
tp/Texinfo/XS/main/document_types.h
(COLLATION_INDICES_SORTED_BY_LETTER)
(COLLATIONS_INDICES_SORTED_BY_LETTER): add sorted_indices_by_letter
as in Perl.
* tp/Texinfo/XS/convert/converter.c
(get_converter_indices_sorted_by_letter): call document
sorted_indices_by_letter. Remove converter_sort_indices_by_letter.
* tp/Texinfo/Convert/Converter.pm
(get_converter_indices_sorted_by_letter),
tp/Texinfo/XS/convert/ConvertXS.xs
(get_converter_indices_sorted_by_letter): return index sorted by
letter instead of caching it in the converter, it is already cached in
the document.
* tp/Texinfo/Document.pm (%XS_structure_overrides),
tp/Texinfo/XS/main/DocumentXS.xs (indices_sort_strings),
tp/Texinfo/XS/main/build_perl_info.c (build_document)
(find_idx_name_entry_number_sv, build_indices_sort_strings)
(build_sorted_indices_by_letter): add XS interface for
indices_sort_strings and sorted_indices_by_letter.
* tp/t/test_utils.pl (test), tp/texi2any.pl: call indices_sort_strings
early such that error messages from XS can be collected.
---
ChangeLog | 66 +++
tp/Texinfo/Convert/Converter.pm | 16 +-
tp/Texinfo/Convert/LaTeX.pm | 3 +
tp/Texinfo/Document.pm | 3 +-
tp/Texinfo/Indices.pm | 7 +-
tp/Texinfo/XS/convert/ConvertXS.xs | 2 +-
tp/Texinfo/XS/convert/converter.c | 35 +-
tp/Texinfo/XS/convert/converter.h | 1 -
tp/Texinfo/XS/main/DocumentXS.xs | 34 ++
tp/Texinfo/XS/main/IndicesXS.xs | 6 +-
tp/Texinfo/XS/main/build_perl_info.c | 216 ++++++--
tp/Texinfo/XS/main/build_perl_info.h | 5 +-
tp/Texinfo/XS/main/call_perl_function.c | 95 ++++
tp/Texinfo/XS/main/call_perl_function.h | 6 +
tp/Texinfo/XS/main/converter_types.h | 12 -
tp/Texinfo/XS/main/document.c | 151 ++++++
tp/Texinfo/XS/main/document.h | 10 +
tp/Texinfo/XS/main/document_types.h | 64 +++
tp/Texinfo/XS/main/manipulate_indices.c | 561 ++++++++++++++++-----
tp/Texinfo/XS/main/manipulate_indices.h | 15 +-
tp/Texinfo/options_data.txt | 4 +
tp/t/test_utils.pl | 7 +-
.../res_parser/formatting_epub/formatting.2 | 2 +-
.../res_parser/formatting_html32/formatting.2 | 2 +-
.../formatting_html_no_split/formatting.2 | 2 +-
.../res_parser/formatting_info/formatting.2 | 2 +-
.../res_parser/formatting_plaintext/formatting.2 | 2 +-
.../res_parser/formatting_xhtml/formatting.2 | 2 +-
.../formatting_enable_encoding/formatting.2 | 2 +-
.../res_parser/formatting_epub_nodes/formatting.2 | 2 +-
.../res_parser/formatting_exotic/formatting.2 | 2 +-
.../layout/res_parser/formatting_fr/formatting.2 | 2 +-
.../res_parser/formatting_fr_icons/formatting.2 | 2 +-
.../res_parser/formatting_fr_info/formatting.2 | 2 +-
.../formatting_info_ascii_punctuation/formatting.2 | 2 +-
.../formatting_info_disable_encoding/formatting.2 | 2 +-
.../res_parser/formatting_inline_css/formatting.2 | 2 +-
.../res_parser/formatting_mathjax/formatting.2 | 2 +-
.../formatting_numerical_entities/formatting.2 | 2 +-
.../formatting.2 | 2 +-
.../formatting_sort_element_counts/formatting.2 | 2 +-
.../res_parser/formatting_texi2html/formatting.2 | 2 +-
.../formatting_texi2html_nodes/formatting.2 | 2 +-
.../formatting_weird_quotes/formatting.2 | 2 +-
.../res_parser/formatting_singular/formatting.2 | 2 +-
tp/texi2any.pl | 9 +
46 files changed, 1101 insertions(+), 273 deletions(-)
diff --git a/ChangeLog b/ChangeLog
index b15334975e..d7fb177a5f 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,69 @@
+2024-02-12 Patrice Dumas <pertusus@free.fr>
+
+ * tp/Texinfo/options_data.txt (DOCUMENTLANGUAGE_COLLATION)
+ (XS_STRXFRM_COLLATION_LOCALE): add customization variables.
+
+ * tp/Texinfo/XS/main/call_perl_function.c (call_setup_collator)
+ (call_collator_getSortKey), tp/Texinfo/XS/main/document_types.h
+ (BYTES_STRING, enum collation_type_name),
+ tp/Texinfo/XS/main/manipulate_indices.c (INDEX_COLLATOR, get_sort_key)
+ (setup_collator, compare_byte_strings, compare_index_letter)
+ (compare_sortable_subentry_keys),
+ tp/Texinfo/XS/main/manipulate_indices.h (SORTABLE_INDEX_SUBENTRY):
+ add an enum for the types of collations, and a structure for a
+ collator. Add the possibility to call perl code to get a collator
+ object and get Perl sort keys as a possibility for collation sort
+ strings. Use bytes strings with length through the BYTES_STRING
+ structure, to be able to use strings with NUL within as collation
+ string keys. This is needed for collation keys coming from Perl and a
+ logical structure in C for collation binary strings. Use memcmp to
+ compare.
+
+ * tp/Texinfo/XS/main/document.c (document_indices_sort_strings)
+ (destroy_document_information_except_tree),
+ tp/Texinfo/XS/main/document_types.h (INDEX_SUBENTRY_SORT_STRING)
+ (INDEX_ENTRY_SORT_STRING, INDEX_SORT_STRINGS, INDICES_SORT_STRINGS)
+ (DOCUMENT), tp/Texinfo/XS/main/manipulate_indices.c
+ (destroy_index_entries_sort_strings),
+ tp/Texinfo/XS/main/manipulate_indices.c
+ (setup_index_entries_sort_strings, setup_sortable_index_entries)
+ (setup_sort_sortable_strings_collator, sort_indices_by_letter): add
+ setup_index_entries_sort_strings based on setup_sortable_index_entries
+ code and add setup_sort_sortable_strings_collator, as in Perl. Add
+ document_indices_sort_strings.
+
+ * tp/Texinfo/XS/main/manipulate_indices.c
+ (destroy_indices_sortable_entries): add.
+
+ * tp/Texinfo/XS/main/document.c (new_collation_sorted_indices)
+ (find_collation_sorted_indices, sorted_indices_by_letter)
+ (destroy_document_information_except_tree),
+ tp/Texinfo/XS/main/document_types.h
+ (COLLATION_INDICES_SORTED_BY_LETTER)
+ (COLLATIONS_INDICES_SORTED_BY_LETTER): add sorted_indices_by_letter
+ as in Perl.
+
+ * tp/Texinfo/XS/convert/converter.c
+ (get_converter_indices_sorted_by_letter): call document
+ sorted_indices_by_letter. Remove converter_sort_indices_by_letter.
+
+ * tp/Texinfo/Convert/Converter.pm
+ (get_converter_indices_sorted_by_letter),
+ tp/Texinfo/XS/convert/ConvertXS.xs
+ (get_converter_indices_sorted_by_letter): return index sorted by
+ letter instead of caching it in the converter, it is already cached in
+ the document.
+
+ * tp/Texinfo/Document.pm (%XS_structure_overrides),
+ tp/Texinfo/XS/main/DocumentXS.xs (indices_sort_strings),
+ tp/Texinfo/XS/main/build_perl_info.c (build_document)
+ (find_idx_name_entry_number_sv, build_indices_sort_strings)
+ (build_sorted_indices_by_letter): add XS interface for
+ indices_sort_strings and sorted_indices_by_letter.
+
+ * tp/t/test_utils.pl (test), tp/texi2any.pl: call indices_sort_strings
+ early such that error messages from XS can be collected.
+
2024-02-11 Gavin Smith <gavinsmith0123@gmail.com>
* doc/texinfo.texi (Other Customization Variables)
diff --git a/tp/Texinfo/Convert/Converter.pm b/tp/Texinfo/Convert/Converter.pm
index 031d659c3c..afd07ca12a 100644
--- a/tp/Texinfo/Convert/Converter.pm
+++ b/tp/Texinfo/Convert/Converter.pm
@@ -1750,15 +1750,8 @@ sub get_converter_indices_sorted_by_letter($)
}
if ($indices_information) {
- if (!$self->get_conf('TEST') and $self->{'converter_descriptor'}
- and $XS_convert) {
- # get from XS
- if ($self->{'index_entries_by_letter'}) {
- return $self->{'index_entries_by_letter'};
- }
-
- $self->{'index_entries_by_letter'}
- = _XS_get_converter_indices_sorted_by_letter($self,
+ if (!$self->{'converter_descriptor'} and $XS_convert) {
+ return _XS_get_converter_indices_sorted_by_letter($self,
$indices_information);
} else {
my $use_unicode_collation
@@ -1768,13 +1761,12 @@ sub get_converter_indices_sorted_by_letter($)
$locale_lang = $self->get_conf('COLLATION_LANGUAGE');
}
- $self->{'index_entries_by_letter'}
- = Texinfo::Document::sorted_indices_by_letter(undef, $self,
+ return Texinfo::Document::sorted_indices_by_letter(undef, $self,
$self->{'document'},
$use_unicode_collation, $locale_lang);
}
}
- return $self->{'index_entries_by_letter'};
+ return {};
}
sub _count_converted_text($$)
diff --git a/tp/Texinfo/Convert/LaTeX.pm b/tp/Texinfo/Convert/LaTeX.pm
index 644cc6948d..0c1069e530 100644
--- a/tp/Texinfo/Convert/LaTeX.pm
+++ b/tp/Texinfo/Convert/LaTeX.pm
@@ -2504,6 +2504,9 @@ sub _index_entry($$)
Texinfo::Convert::Text::set_options_code(
$self->{'index_formatting_text_options'});
}
+ # NOTE in XS code, the $self->{'index_formatting_text_options'}
+ # argument is ignored, instead the $self converter is used to find C
+ # text options data setup by setup_index_entry_keys_formatting.
my $sort_string
= Texinfo::Indices::index_entry_element_sort_string(
$self, $entry,
diff --git a/tp/Texinfo/Document.pm b/tp/Texinfo/Document.pm
index 69f230a371..269a3ecabc 100644
--- a/tp/Texinfo/Document.pm
+++ b/tp/Texinfo/Document.pm
@@ -60,6 +60,8 @@ our %XS_structure_overrides = (
=> "Texinfo::DocumentXS::rebuild_document",
"Texinfo::Document::rebuild_tree"
=> "Texinfo::DocumentXS::rebuild_tree",
+ "Texinfo::Document::indices_sort_strings"
+ => "Texinfo::DocumentXS::indices_sort_strings",
);
our $module_loaded = 0;
@@ -206,7 +208,6 @@ sub merged_indices($)
# TODO document
# call setup_index_entries_sort_strings and cache the result.
-# TODO XS override
sub indices_sort_strings($$$;$)
{
my $registrar = shift;
diff --git a/tp/Texinfo/Indices.pm b/tp/Texinfo/Indices.pm
index 0b34434f25..9b524ddf87 100644
--- a/tp/Texinfo/Indices.pm
+++ b/tp/Texinfo/Indices.pm
@@ -333,7 +333,7 @@ sub setup_index_entries_sort_strings($$$$;$)
Texinfo::Convert::Text::set_options_code($convert_text_options);
}
my $entry_sort_string
- = index_entry_element_sort_string ($customization_information,
+ = index_entry_element_sort_string($customization_information,
$index_entry, $main_entry_element,
$convert_text_options, $prefer_reference_element);
my $non_empty_index_subentries = 0;
@@ -476,6 +476,8 @@ sub _setup_sort_sortable_strings_collator($$$$;$$$)
my $indices_sort_strings;
if ($document) {
+ # simple wrapper around setup_index_entries_sort_strings that caches the
+ # result
$indices_sort_strings = Texinfo::Document::indices_sort_strings($registrar,
$customization_information, $document);
} else {
@@ -638,7 +640,7 @@ sub index_entry_first_letter_text_or_command($;$)
}
}
-sub sort_indices_by_letter($$$$$$;$)
+sub sort_indices_by_letter($$$$;$$$)
{
my $registrar = shift;
my $customization_information = shift;
@@ -665,7 +667,6 @@ sub sort_indices_by_letter($$$$$$;$)
foreach my $sortable_entry (@{$sortable_index_entries}) {
my $entry_key
= $sortable_entry->{'entry_strings_alpha'}->[0]->{'sort_string'};
-
# the following line leads to each accented letter being separate
# $letter = uc(substr($entry_key, 0, 1));
my $letter_string = uc(substr($entry_key, 0, 1));
diff --git a/tp/Texinfo/XS/convert/ConvertXS.xs
b/tp/Texinfo/XS/convert/ConvertXS.xs
index 7bc035a4ba..e8b91de8dd 100644
--- a/tp/Texinfo/XS/convert/ConvertXS.xs
+++ b/tp/Texinfo/XS/convert/ConvertXS.xs
@@ -190,7 +190,7 @@ get_converter_indices_sorted_by_letter (SV *converter_sv,
SV *indices_informatio
if (self)
{
INDEX_SORTED_BY_LETTER *index_entries_by_letter
- = converter_sort_indices_by_letter (self);
+ = get_converter_indices_sorted_by_letter (self);
RETVAL
= build_sorted_indices_by_letter (index_entries_by_letter,
indices_information);
diff --git a/tp/Texinfo/XS/convert/converter.c
b/tp/Texinfo/XS/convert/converter.c
index 59373fa38e..3c7285d1f9 100644
--- a/tp/Texinfo/XS/convert/converter.c
+++ b/tp/Texinfo/XS/convert/converter.c
@@ -637,39 +637,18 @@ free_comma_index_subentries_tree (ELEMENT_LIST
*element_list)
destroy_list (element_list);
}
-INDEX_SORTED_BY_LETTER *
-converter_sort_indices_by_letter (CONVERTER *self)
-{
- if (self->index_entries_by_letter)
- return self->index_entries_by_letter;
-
- const MERGED_INDICES *merged_indices
- = document_merged_indices (self->document);
-
- self->index_entries_by_letter
- = sort_indices_by_letter (&self->error_messages, self->conf,
- merged_indices,
- self->document->index_names);
- return self->index_entries_by_letter;
-}
-
INDEX_SORTED_BY_LETTER *
get_converter_indices_sorted_by_letter (CONVERTER *self)
{
- if (self->index_entries_by_letter)
- return self->index_entries_by_letter;
-
- if (self->document->index_names)
+ if (self->document)
{
- /* get Perl sorting for reproducible tests */
- if (self->conf->TEST.integer > 0)
- self->index_entries_by_letter
- = get_call_index_entries_sorted_by_letter (self);
- else /* sets self->index_entries_by_letter */
- converter_sort_indices_by_letter (self);
+ return sorted_indices_by_letter (&self->error_messages, self->conf,
+ self->document,
+ self->conf->USE_UNICODE_COLLATION.integer,
+ self->conf->COLLATION_LANGUAGE.string,
+ self->conf->XS_STRXFRM_COLLATION_LOCALE.string);
}
-
- return self->index_entries_by_letter;
+ return 0;
}
/* to be freed by caller */
diff --git a/tp/Texinfo/XS/convert/converter.h
b/tp/Texinfo/XS/convert/converter.h
index 63170a55fa..66915297e8 100644
--- a/tp/Texinfo/XS/convert/converter.h
+++ b/tp/Texinfo/XS/convert/converter.h
@@ -114,7 +114,6 @@ ELEMENT_LIST *comma_index_subentries_tree (const ELEMENT
*current_entry,
char *separator);
void free_comma_index_subentries_tree (ELEMENT_LIST *element);
-INDEX_SORTED_BY_LETTER *converter_sort_indices_by_letter (CONVERTER *self);
INDEX_SORTED_BY_LETTER *get_converter_indices_sorted_by_letter
(CONVERTER *self);
diff --git a/tp/Texinfo/XS/main/DocumentXS.xs b/tp/Texinfo/XS/main/DocumentXS.xs
index bed412bef4..3fc4e4e923 100644
--- a/tp/Texinfo/XS/main/DocumentXS.xs
+++ b/tp/Texinfo/XS/main/DocumentXS.xs
@@ -176,6 +176,40 @@ set_document_options (SV *sv_options_in, SV *document_in)
register_document_options (document, options);
}
+SV *
+indices_sort_strings (SV *registrar, SV *main_configuration, SV *document_in,
int prefer_reference_element=0)
+ PREINIT:
+ DOCUMENT *document = 0;
+ const INDICES_SORT_STRINGS *indices_sort_strings = 0;
+ SV **indices_information_sv;
+ HV *document_hv;
+ CODE:
+ document = get_sv_document_document (document_in,
+ "indices_sort_strings");
+ if (document)
+ indices_sort_strings
+ = document_indices_sort_strings (document->error_messages,
+ document->options, document,
+ prefer_reference_element);
+
+ document_hv = (HV *) SvRV (document_in);
+ indices_information_sv
+ = hv_fetch (document_hv, "indices", strlen ("indices"), 0);
+
+ if (indices_sort_strings && indices_information_sv)
+ {
+ HV *indices_information_hv = (HV *) SvRV (*indices_information_sv);
+ HV *indices_sort_strings_hv
+ = build_indices_sort_strings (indices_sort_strings,
+ indices_information_hv);
+
+ RETVAL = newRV_inc ((SV *) indices_sort_strings_hv);
+ }
+ else
+ RETVAL = newSV (0);
+ OUTPUT:
+ RETVAL
+
# Next correspond to XS interfaces that have no associated
# .xs file.
diff --git a/tp/Texinfo/XS/main/IndicesXS.xs b/tp/Texinfo/XS/main/IndicesXS.xs
index 5cd20fb47d..e863ae2ea5 100644
--- a/tp/Texinfo/XS/main/IndicesXS.xs
+++ b/tp/Texinfo/XS/main/IndicesXS.xs
@@ -47,13 +47,13 @@ MODULE = Texinfo::IndicesXS PACKAGE = Texinfo::IndicesXS
PROTOTYPES: ENABLE
# This function triggers setting the information needed for calls
-# to index_entry_element_sort_string in C, either in a document
+# to index_entry_element_sort_string through XS, either in a document
# or in a converter, depending whether index sorting is done with a converter
# or without.
# A returned hash reference is needed as some information is set in the
-# hash in perl afterwards, but this information is not used as the hash
+# hash in Perl afterwards, but this information is not used as the hash
# is not used by any function not overriden, so there is no need to
-# return information corresponding to the text options for perl. An empty
+# return information corresponding to the text options for Perl. An empty
# hash reference is therefore returned.
SV *
setup_index_entry_keys_formatting (SV *customization_info_sv)
diff --git a/tp/Texinfo/XS/main/build_perl_info.c
b/tp/Texinfo/XS/main/build_perl_info.c
index 9ced638642..d53e17b64f 100644
--- a/tp/Texinfo/XS/main/build_perl_info.c
+++ b/tp/Texinfo/XS/main/build_perl_info.c
@@ -1216,6 +1216,7 @@ build_document (size_t document_descriptor, int no_store)
HV *hv_commands_info;
HV *hv_index_names;
HV *hv_listoffloats_list;
+ HV *hv_indices_sort_strings = 0;
AV *av_internal_xref;
HV *hv_identifiers_target;
AV *av_labels_list;
@@ -1267,6 +1268,11 @@ build_document (size_t document_descriptor, int no_store)
if (document->sections_list)
av_sections_list = build_elements_list (document->sections_list);
+ if (document->indices_sort_strings)
+ hv_indices_sort_strings = build_indices_sort_strings (
+ document->indices_sort_strings,
+ hv_index_names);
+
#define STORE(key, value) hv_store (hv, key, strlen (key), newRV_inc ((SV *)
value), 0)
/* must be kept in sync with Texinfo::Document register keys */
@@ -1285,6 +1291,9 @@ build_document (size_t document_descriptor, int no_store)
if (av_sections_list)
STORE("sections_list", av_sections_list);
+
+ if (hv_indices_sort_strings)
+ STORE("index_entries_sort_strings", hv_indices_sort_strings);
#undef STORE
if (no_store)
@@ -1922,14 +1931,162 @@ build_expanded_formats (EXPANDED_FORMAT
*expanded_formats)
return newRV_noinc ((SV *)expanded_hv);
}
+SV *
+find_idx_name_entry_number_sv (HV *indices_information_hv,
+ const char* index_name, int entry_number,
+ const char *message)
+{
+ SV **index_info_sv;
+ SV *index_entry_sv = 0;
+
+ dTHX;
+
+ index_info_sv = hv_fetch (indices_information_hv, index_name,
+ strlen (index_name), 0);
+ if (!index_info_sv)
+ {
+ fprintf (stderr, "%s index %s not found\n", message, index_name);
+ }
+ else
+ {
+ HV *index_info_hv = (HV *) SvRV (*index_info_sv);
+ SV **index_info_index_entries_sv = hv_fetch (index_info_hv,
+ "index_entries", strlen ("index_entries"), 0);
+
+ if (!index_info_index_entries_sv)
+ {
+ fprintf (stderr, "%s index %s 'index_entries' not found\n",
+ message, index_name);
+ }
+ else
+ {
+ AV *index_info_entries_av
+ = (AV *) SvRV (*index_info_index_entries_sv);
+
+ SV **index_entry_info_sv = av_fetch (index_info_entries_av,
+ entry_number -1, 0);
+
+ if (!index_entry_info_sv)
+ {
+ fprintf (stderr, "%s: %d in %s not found\n", message,
+ entry_number, index_name);
+ }
+ else
+ index_entry_sv = *index_entry_info_sv;
+ }
+ }
+ return index_entry_sv;
+}
+
+HV *
+build_indices_sort_strings (const INDICES_SORT_STRINGS *indices_sort_strings,
+ HV *indices_information_hv)
+{
+ HV *indices_sort_strings_hv;
+ size_t i;
+
+ dTHX;
+
+ if (!indices_sort_strings)
+ return 0;
+
+ indices_sort_strings_hv = newHV ();
+
+ for (i = 0; i < indices_sort_strings->number; i++)
+ {
+ INDEX_SORT_STRINGS *index_sort_strings
+ = &indices_sort_strings->indices[i];
+ char *index_name = index_sort_strings->index->name;
+
+ if (index_sort_strings->entries_number > 0)
+ {
+ size_t j;
+ AV *sort_string_entries_av = newAV ();
+
+ hv_store (indices_sort_strings_hv, index_name, strlen (index_name),
+ newRV_noinc ((SV *)sort_string_entries_av), 0);
+
+ for (j = 0; j < index_sort_strings->entries_number; j++)
+ {
+ INDEX_ENTRY_SORT_STRING *index_entry_sort_string
+ = &index_sort_strings->sort_string_entries[j];
+ INDEX_ENTRY *entry = index_entry_sort_string->entry;
+ char *entry_index_name = entry->index_name;
+ int entry_number = entry->number;
+ char *message;
+ SV *index_entry_sv;
+ HV *index_entry_sort_string_hv;
+ AV *sort_string_subentries_av;
+ size_t k;
+
+ if (index_entry_sort_string->subentries_number <= 0)
+ {
+ fprintf (stderr, "BUG: build_indices_sort_strings:"
+ " %s: entry %zu: no subentries", index_name, j);
+ continue;
+ }
+
+ xasprintf (&message, "BUG: build_indices_sort_strings:"
+ " %s: entry %zu", index_name, j);
+ index_entry_sv
+ = find_idx_name_entry_number_sv (indices_information_hv,
+ entry_index_name,
entry_number,
+ message);
+ free (message);
+
+ /* probably not possible, unless there is a bug */
+ if (!index_entry_sv)
+ continue;
+
+ index_entry_sort_string_hv = newHV ();
+ av_push (sort_string_entries_av,
+ newRV_noinc ((SV *) index_entry_sort_string_hv));
+
+ hv_store (index_entry_sort_string_hv, "index_name",
+ strlen ("index_name"),
+ newSVpv_utf8 (entry->index_name, 0), 0);
+ hv_store (index_entry_sort_string_hv, "number",
+ strlen ("number"), newSViv (entry->number), 0);
+
+ SvREFCNT_inc (index_entry_sv);
+ hv_store (index_entry_sort_string_hv, "entry",
+ strlen ("entry"), index_entry_sv, 0);
+
+ sort_string_subentries_av = newAV ();
+ hv_store (index_entry_sort_string_hv, "sort_strings",
+ strlen ("sort_strings"),
+ newRV_noinc ((SV *) sort_string_subentries_av), 0);
+
+ for (k = 0; k < index_entry_sort_string->subentries_number; k++)
+ {
+ INDEX_SUBENTRY_SORT_STRING *subentry_sort_string
+ = &index_entry_sort_string->sort_string_subentries[k];
+ HV *subentry_sort_string_hv = newHV ();
+
+ av_push (sort_string_subentries_av,
+ newRV_noinc ((SV *) subentry_sort_string_hv));
+
+ hv_store (subentry_sort_string_hv, "sort_string",
+ strlen ("sort_string"),
+ newSVpv_utf8 (subentry_sort_string->sort_string, 0), 0);
+ hv_store (subentry_sort_string_hv, "alpha",
+ strlen ("alpha"),
+ newSViv (subentry_sort_string->alpha), 0);
+ }
+ }
+ }
+ }
+ return indices_sort_strings_hv;
+}
+
SV *
build_sorted_indices_by_letter (
- INDEX_SORTED_BY_LETTER *index_entries_by_letter,
+ const INDEX_SORTED_BY_LETTER *index_entries_by_letter,
SV *indices_information)
{
HV *indices_hv;
HV *indices_information_hv;
- INDEX_SORTED_BY_LETTER *idx;
+ const INDEX_SORTED_BY_LETTER *idx;
dTHX;
@@ -1973,51 +2130,16 @@ build_sorted_indices_by_letter (
INDEX_ENTRY *entry = letter->entries[j];
char *index_name = entry->index_name;
int entry_number = entry->number;
- SV **index_info_sv;
- SV *index_entry_sv = 0;
-
- index_info_sv = hv_fetch (indices_information_hv, index_name,
- strlen (index_name), 0);
- if (!index_info_sv)
- {
- fprintf (stderr,
- "BUG: build_sorted_indices_by_letter: "
- "%s: %s: entry %zu index %s not found\n",
- idx->name, letter->letter, j, index_name);
- }
- else
- {
- HV *index_info_hv = (HV *) SvRV (*index_info_sv);
- SV **index_info_index_entries_sv = hv_fetch (index_info_hv,
- "index_entries", strlen ("index_entries"), 0);
-
- if (!index_info_index_entries_sv)
- {
- fprintf (stderr,
- "BUG: build_sorted_indices_by_letter: "
- "%s: %s: entry %zu index %s 'index_entries' not found\n",
- idx->name, letter->letter, j, index_name);
- }
- else
- {
- AV *index_info_entries_av
- = (AV *) SvRV (*index_info_index_entries_sv);
-
- SV **index_entry_info_sv = av_fetch
(index_info_entries_av,
- entry_number -1, 0);
-
- if (!index_entry_info_sv)
- {
- fprintf (stderr,
- "BUG: build_sorted_indices_by_letter: "
- "%s: %s: entry %zu: %d in %s not found\n",
- idx->name, letter->letter, j,
- entry_number, index_name);
- }
- else
- index_entry_sv = *index_entry_info_sv;
- }
- }
+ char *message;
+ SV *index_entry_sv;
+ xasprintf (&message, "BUG: build_sorted_indices_by_letter:"
+ " %s: %s: entry %zu", idx->name,
+ letter->letter, j);
+ index_entry_sv
+ = find_idx_name_entry_number_sv (indices_information_hv,
+ index_name, entry_number,
+ message);
+ free (message);
if (index_entry_sv)
{
diff --git a/tp/Texinfo/XS/main/build_perl_info.h
b/tp/Texinfo/XS/main/build_perl_info.h
index 5941261d79..9cf6baed78 100644
--- a/tp/Texinfo/XS/main/build_perl_info.h
+++ b/tp/Texinfo/XS/main/build_perl_info.h
@@ -53,8 +53,11 @@ void pass_output_unit_files (SV *converter_sv,
void build_output_files_information (SV *converter_sv,
OUTPUT_FILES_INFORMATION *output_files_information);
+HV *build_indices_sort_strings (
+ const INDICES_SORT_STRINGS *indices_sort_strings,
+ HV *indices_information_hv);
SV *build_sorted_indices_by_letter (
- INDEX_SORTED_BY_LETTER *index_entries_by_letter,
+ const INDEX_SORTED_BY_LETTER *index_entries_by_letter,
SV *indices_information);
SV *html_build_direction_icons (CONVERTER *converter,
diff --git a/tp/Texinfo/XS/main/call_perl_function.c
b/tp/Texinfo/XS/main/call_perl_function.c
index b270bff09a..7ddc0c8ef0 100644
--- a/tp/Texinfo/XS/main/call_perl_function.c
+++ b/tp/Texinfo/XS/main/call_perl_function.c
@@ -34,6 +34,7 @@
#include "tree_types.h"
#include "converter_types.h"
+#include "document_types.h"
#include "build_perl_info.h"
/* for get_sv_index_entries_sorted_by_letter */
#include "get_perl_info.h"
@@ -208,3 +209,97 @@ get_call_index_entries_sorted_by_letter (CONVERTER *self)
return result;
}
+const void *
+call_setup_collator (int use_unicode_collation, const char *locale_lang)
+{
+ int count;
+ const void *result = 0;
+ SV *collator_sv = 0;
+
+ dTHX;
+
+ dSP;
+
+ ENTER;
+ SAVETMPS;
+
+ PUSHMARK(SP);
+ EXTEND(SP, 2);
+
+ PUSHs(sv_2mortal (newSViv (use_unicode_collation)));
+ PUSHs(sv_2mortal (newSVpv (locale_lang, 0)));
+
+ PUTBACK;
+
+ count = call_pv ("Texinfo::Indices::_setup_collator",
+ G_SCALAR);
+
+ SPAGAIN;
+
+ if (count != 1)
+ croak("_setup_collator should return 1 item\n");
+
+ collator_sv = POPs;
+ if (SvOK (collator_sv))
+ {
+ SvREFCNT_inc (collator_sv);
+ result = (const void *) collator_sv;
+ }
+
+ PUTBACK;
+
+ FREETMPS;
+ LEAVE;
+
+ return result;
+}
+
+BYTES_STRING *
+call_collator_getSortKey (const void *collator_sv, const char *string)
+{
+ BYTES_STRING *result;
+ unsigned char *result_ret;
+ STRLEN len;
+ SV *result_sv;
+ int count;
+
+ dTHX;
+
+ result = malloc (sizeof (BYTES_STRING));
+ dSP;
+
+ ENTER;
+ SAVETMPS;
+
+ PUSHMARK(SP);
+ EXTEND(SP, 2);
+
+ PUSHs((SV *) collator_sv);
+ PUSHs(sv_2mortal (newSVpv_utf8 (string, 0)));
+
+ PUTBACK;
+
+ count = call_method ("getSortKey",
+ G_SCALAR);
+
+ SPAGAIN;
+
+ if (count != 1)
+ croak("getSortKey should return 1 item\n");
+
+ result_sv = POPs;
+ result_ret = (unsigned char *)SvPVbyte (result_sv, len);
+ result->len = (size_t) len;
+ result->bytes = (unsigned char *)
+ malloc (sizeof (unsigned char) * len);
+ memcpy (result->bytes, result_ret, result->len);
+
+ PUTBACK;
+
+ FREETMPS;
+ LEAVE;
+
+ return result;
+}
+
+
diff --git a/tp/Texinfo/XS/main/call_perl_function.h
b/tp/Texinfo/XS/main/call_perl_function.h
index cbeb404915..1aad256ea0 100644
--- a/tp/Texinfo/XS/main/call_perl_function.h
+++ b/tp/Texinfo/XS/main/call_perl_function.h
@@ -9,6 +9,12 @@ char *call_nodenamenormalization_unicode_to_transliterate
(char *text);
char *call_latex_convert_to_latex_math (CONVERTER *self, ELEMENT *element);
+const void *call_setup_collator (int use_unicode_collation,
+ const char *locale_lang);
+
+BYTES_STRING *call_collator_getSortKey (const void *collator_sv,
+ const char *string);
+
INDEX_SORTED_BY_LETTER *get_call_index_entries_sorted_by_letter
(CONVERTER *self);
#endif
diff --git a/tp/Texinfo/XS/main/converter_types.h
b/tp/Texinfo/XS/main/converter_types.h
index a4b3f7084a..b475f85fe3 100644
--- a/tp/Texinfo/XS/main/converter_types.h
+++ b/tp/Texinfo/XS/main/converter_types.h
@@ -317,18 +317,6 @@ typedef struct HTML_SHARED_CONVERSION_STATE {
/* formatted_nodedescriptions */
} HTML_SHARED_CONVERSION_STATE;
-typedef struct LETTER_INDEX_ENTRIES {
- char *letter;
- INDEX_ENTRY **entries;
- size_t entries_number;
-} LETTER_INDEX_ENTRIES;
-
-typedef struct INDEX_SORTED_BY_LETTER {
- char *name;
- LETTER_INDEX_ENTRIES *letter_entries;
- size_t letter_number;
-} INDEX_SORTED_BY_LETTER;
-
typedef struct HTML_COMMAND_CONVERSION {
char *element;
int quote; /* for style commands formatting only */
diff --git a/tp/Texinfo/XS/main/document.c b/tp/Texinfo/XS/main/document.c
index b1c58ae64e..c082a2695e 100644
--- a/tp/Texinfo/XS/main/document.c
+++ b/tp/Texinfo/XS/main/document.c
@@ -157,6 +157,137 @@ register_document_convert_index_text_options (DOCUMENT
*document,
document->convert_index_text_options = text_options;
}
+const INDICES_SORT_STRINGS *
+document_indices_sort_strings (ERROR_MESSAGE_LIST *error_messages,
+ OPTIONS *options, DOCUMENT *document,
+ int prefer_reference_element)
+{
+ if (!document->indices_sort_strings)
+ {
+ const MERGED_INDICES *merged_indices
+ = document_merged_indices (document);
+
+ document->indices_sort_strings
+ = setup_index_entries_sort_strings (error_messages, options,
+ merged_indices, document->index_names,
+ prefer_reference_element);
+ }
+
+ return document->indices_sort_strings;
+}
+
+static COLLATION_INDICES_SORTED_BY_LETTER *
+new_collation_sorted_indices (
+ COLLATIONS_INDICES_SORTED_BY_LETTER *collations,
+ enum collation_type_name type,
+ const char *language)
+{
+ COLLATION_INDICES_SORTED_BY_LETTER *result = 0;
+ if (collations->number <= collations->space)
+ {
+ collations->collation_sorted_indices
+ = (COLLATION_INDICES_SORTED_BY_LETTER *) realloc
+ (collations->collation_sorted_indices,
+ (collations->space += 3)
+ * sizeof (COLLATION_INDICES_SORTED_BY_LETTER));
+ if (!collations->collation_sorted_indices)
+ fatal ("realloc failed");
+ }
+
+ result = &collations->collation_sorted_indices[collations->number];
+ memset (result, 0, sizeof (COLLATION_INDICES_SORTED_BY_LETTER));
+ result->type = type;
+ result->language = strdup (language);
+
+ collations->number++;
+
+ return result;
+}
+
+COLLATION_INDICES_SORTED_BY_LETTER *
+find_collation_sorted_indices (
+ COLLATIONS_INDICES_SORTED_BY_LETTER *collations,
+ enum collation_type_name type,
+ const char *language)
+{
+ size_t i;
+ for (i = 2; i < collations->number; i++)
+ {
+ COLLATION_INDICES_SORTED_BY_LETTER *collation_sorted_indices
+ = &collations->collation_sorted_indices[i];
+ if (collation_sorted_indices->type == type
+ && !strcmp (collation_sorted_indices->language, language))
+ return collation_sorted_indices;
+ }
+ return 0;
+}
+
+INDEX_SORTED_BY_LETTER *
+sorted_indices_by_letter (ERROR_MESSAGE_LIST *error_messages,
+ OPTIONS *options, DOCUMENT *document,
+ int use_unicode_collation,
+ const char *collation_language,
+ const char *collation_locale)
+{
+ COLLATIONS_INDICES_SORTED_BY_LETTER *collations;
+ COLLATION_INDICES_SORTED_BY_LETTER *collation_sorted_indices = 0;
+ if (!document->sorted_indices_by_letter)
+ {
+ collations
+ = (COLLATIONS_INDICES_SORTED_BY_LETTER *)
+ malloc (sizeof (COLLATIONS_INDICES_SORTED_BY_LETTER));
+ memset (collations, 0,
+ sizeof (COLLATIONS_INDICES_SORTED_BY_LETTER));
+
+ /* order is important, to match enum */
+ new_collation_sorted_indices (collations, ctn_unicode, "-");
+ new_collation_sorted_indices (collations, ctn_no_unicode, "");
+
+ document->sorted_indices_by_letter = collations;
+ }
+
+ collations = document->sorted_indices_by_letter;
+
+ if (use_unicode_collation == 0)
+ collation_sorted_indices
+ = &collations->collation_sorted_indices[ctn_no_unicode];
+ else if (!collation_language && !collation_locale)
+ collation_sorted_indices
+ = &collations->collation_sorted_indices[ctn_unicode];
+ else
+ {
+ enum collation_type_name type;
+ const char *language;
+ if (collation_language)
+ {
+ type = ctn_language_collation;
+ language = collation_language;
+ }
+ else
+ {
+ type = ctn_locale_collation;
+ language = collation_locale;
+ }
+ collation_sorted_indices
+ = find_collation_sorted_indices (collations, type, language);
+ if (!collation_sorted_indices)
+ collation_sorted_indices = new_collation_sorted_indices (collations,
+ type, language);
+ }
+
+ if (!collation_sorted_indices->sorted_indices)
+ {
+ const MERGED_INDICES *merged_indices
+ = document_merged_indices (document);
+ collation_sorted_indices->sorted_indices
+ = sort_indices_by_letter (error_messages, options,
+ use_unicode_collation, collation_language,
+ collation_locale, merged_indices,
+ document->index_names, document);
+ }
+ return collation_sorted_indices->sorted_indices;
+}
+
void
destroy_document_information_except_tree (DOCUMENT *document)
{
@@ -191,6 +322,26 @@ destroy_document_information_except_tree (DOCUMENT
*document)
destroy_text_options (document->convert_index_text_options);
if (document->merged_indices)
destroy_merged_indices (document->merged_indices);
+ if (document->indices_sort_strings)
+ destroy_index_entries_sort_strings (document->indices_sort_strings);
+ if (document->sorted_indices_by_letter)
+ {
+ if (document->sorted_indices_by_letter->number > 0)
+ {
+ size_t i;
+ for (i = 0; i < document->sorted_indices_by_letter->number; i++)
+ {
+ COLLATION_INDICES_SORTED_BY_LETTER *collation_sorted_indices
+ = &document->sorted_indices_by_letter->collation_sorted_indices[i];
+ free (collation_sorted_indices->language);
+ if (collation_sorted_indices->sorted_indices)
+ destroy_indices_sorted_by_letter (
+ collation_sorted_indices->sorted_indices);
+ }
+ }
+ free (document->sorted_indices_by_letter->collation_sorted_indices);
+ free (document->sorted_indices_by_letter);
+ }
}
}
diff --git a/tp/Texinfo/XS/main/document.h b/tp/Texinfo/XS/main/document.h
index 3359b20c41..2682dd7e63 100644
--- a/tp/Texinfo/XS/main/document.h
+++ b/tp/Texinfo/XS/main/document.h
@@ -31,6 +31,16 @@ void register_document_convert_index_text_options (DOCUMENT
*document,
struct TEXT_OPTIONS *text_options);
const MERGED_INDICES *document_merged_indices (DOCUMENT *document);
+const INDICES_SORT_STRINGS *document_indices_sort_strings (
+ ERROR_MESSAGE_LIST *error_messages,
+ OPTIONS *options, DOCUMENT *document,
+ int prefer_reference_element);
+INDEX_SORTED_BY_LETTER *sorted_indices_by_letter (
+ ERROR_MESSAGE_LIST *error_messages,
+ OPTIONS *options, DOCUMENT *document,
+ int use_unicode_collation,
+ const char *collation_language,
+ const char *collation_locale);
void remove_document_descriptor (int document_descriptor);
ELEMENT *unregister_document_merge_with_document (int document_descriptor,
diff --git a/tp/Texinfo/XS/main/document_types.h
b/tp/Texinfo/XS/main/document_types.h
index 73c1b37fe7..7cd1c2ff4b 100644
--- a/tp/Texinfo/XS/main/document_types.h
+++ b/tp/Texinfo/XS/main/document_types.h
@@ -77,6 +77,68 @@ typedef struct MERGED_INDICES {
MERGED_INDEX *indices;
} MERGED_INDICES;
+/* not used in document, but used for indices for sort keys that can
+ contain NUL */
+typedef struct BYTES_STRING {
+ size_t len;
+ unsigned char *bytes;
+} BYTES_STRING;
+
+typedef struct INDEX_SUBENTRY_SORT_STRING {
+ char *sort_string;
+ int alpha;
+} INDEX_SUBENTRY_SORT_STRING;
+
+typedef struct INDEX_ENTRY_SORT_STRING {
+ INDEX_ENTRY *entry;
+ /* in perl 'index_name' => $index_entry->{'index_name'} */
+ /* in perl 'number' => $index_entry->{'entry_number'} */
+ size_t subentries_number;
+ INDEX_SUBENTRY_SORT_STRING *sort_string_subentries;
+} INDEX_ENTRY_SORT_STRING;
+
+typedef struct INDEX_SORT_STRINGS {
+ MERGED_INDEX *index;
+ size_t entries_number;
+ INDEX_ENTRY_SORT_STRING *sort_string_entries;
+} INDEX_SORT_STRINGS;
+
+typedef struct INDICES_SORT_STRINGS {
+ size_t number;
+ INDEX_SORT_STRINGS *indices;
+} INDICES_SORT_STRINGS;
+
+typedef struct LETTER_INDEX_ENTRIES {
+ char *letter;
+ INDEX_ENTRY **entries;
+ size_t entries_number;
+} LETTER_INDEX_ENTRIES;
+
+typedef struct INDEX_SORTED_BY_LETTER {
+ char *name;
+ LETTER_INDEX_ENTRIES *letter_entries;
+ size_t letter_number;
+} INDEX_SORTED_BY_LETTER;
+
+enum collation_type_name {
+ ctn_unicode, /* the default */
+ ctn_no_unicode,
+ ctn_language_collation,
+ ctn_locale_collation, /* experimental, to test strxfrm */
+};
+
+typedef struct COLLATION_INDICES_SORTED_BY_LETTER {
+ enum collation_type_name type;
+ char *language;
+ INDEX_SORTED_BY_LETTER *sorted_indices;
+} COLLATION_INDICES_SORTED_BY_LETTER;
+
+typedef struct COLLATIONS_INDICES_SORTED_BY_LETTER {
+ size_t number;
+ size_t space;
+ COLLATION_INDICES_SORTED_BY_LETTER *collation_sorted_indices;
+} COLLATIONS_INDICES_SORTED_BY_LETTER;
+
typedef struct DOCUMENT {
int descriptor;
ELEMENT *tree;
@@ -96,6 +158,8 @@ typedef struct DOCUMENT {
struct OPTIONS *options; /* for options used in structuring */
struct TEXT_OPTIONS *convert_index_text_options; /* for index
sorting without converter */
+ INDICES_SORT_STRINGS *indices_sort_strings;
+ COLLATIONS_INDICES_SORTED_BY_LETTER *sorted_indices_by_letter;
} DOCUMENT;
/* not in document, but used in parser */
diff --git a/tp/Texinfo/XS/main/manipulate_indices.c
b/tp/Texinfo/XS/main/manipulate_indices.c
index 1296907f6c..450029a18c 100644
--- a/tp/Texinfo/XS/main/manipulate_indices.c
+++ b/tp/Texinfo/XS/main/manipulate_indices.c
@@ -34,6 +34,7 @@
#include "unicode.h"
#include "convert_to_text.h"
#include "convert_to_texinfo.h"
+#include "call_perl_function.h"
#include "manipulate_indices.h"
/* corresponding perl code in Texinfo::Indices */
@@ -253,62 +254,121 @@ index_entry_element_sort_string (const INDEX_ENTRY
*main_entry,
return sort_string;
}
-static void
-set_sort_key (locale_t collation_locale, const char *input_string,
- char **result_key)
+typedef struct INDEX_COLLATOR {
+ enum collation_type_name type;
+ char *language;
+ union {
+ /* perl element. This should be SV *sv,
+ but we don't want to include the Perl headers everywhere; */
+ const void *sv;
+ #ifdef HAVE_NEWLOCALE
+ locale_t locale;
+ #endif
+ };
+} INDEX_COLLATOR;
+
+static BYTES_STRING *get_sort_key (INDEX_COLLATOR *collator,
+ const char *sort_string)
{
- if (collation_locale)
+ BYTES_STRING *sort_key;
+ switch (collator->type)
{
- #ifdef HAVE_STRXFRM_L
- size_t len = strxfrm_l (0, input_string, 0, collation_locale);
- size_t check_len;
-
- *result_key
- = (char *) malloc ((len +1) * sizeof (char));
- check_len = strxfrm_l (*result_key, input_string, len+1,
- collation_locale);
- if (check_len != len)
- fatal ("strxfrm_l returns a different length");
- #endif
+ case ctn_no_unicode:
+ sort_key = (BYTES_STRING *) malloc (sizeof (BYTES_STRING));
+ sort_key->len = strlen (sort_string);
+ sort_key->bytes = (unsigned char *)
+ malloc (sizeof (unsigned char) * sort_key->len);
+ memcpy (sort_key->bytes, (unsigned char *) sort_string, sort_key->len);
+ break;
+ #ifdef HAVE_STRXFRM_L
+ case ctn_locale_collation:
+ {
+ size_t check_len;
+ char *char_sort_key;
+ sort_key = (BYTES_STRING *) malloc (sizeof (BYTES_STRING));
+ sort_key->len
+ = strxfrm_l (0, sort_string, 0,
+ collator->locale);
+ char_sort_key = (char *) malloc (sizeof (char) * sort_key->len);
+ check_len
+ = strxfrm_l (char_sort_key, sort_string, sort_key->len,
+ collator->locale);
+ sort_key->bytes = (unsigned char *)
+ malloc (sizeof (unsigned char) * sort_key->len);
+ memcpy (sort_key->bytes, (unsigned char *) char_sort_key,
+ sort_key->len);
+ free (char_sort_key);
+ if (check_len != sort_key->len)
+ fatal ("strxfrm_l returns a different length");
+ }
+ break;
+ #endif
+ case ctn_unicode:
+ case ctn_language_collation:
+ default: /* !HAVE_STRXFRM_L && ctn_locale_collation */
+ {
+ char *uc_sort_string
+ = to_upper_or_lower_multibyte (sort_string, 1);
+ sort_key = call_collator_getSortKey (collator->sv,
+ uc_sort_string);
+ free (uc_sort_string);
+ }
+ break;
}
- else
- *result_key = strdup (input_string);
+ return sort_key;
}
-typedef struct INDEX_SORT_STRING_KEY {
- char *sort_string;
- char *sort_key;
-} INDEX_SORT_STRING_KEY;
-
-static INDEX_SORT_STRING_KEY *
-index_entry_element_sort_string_key (const INDEX_ENTRY *main_entry,
- const ELEMENT *index_entry_element,
- TEXT_OPTIONS *options, int in_code,
- locale_t collation_locale,
- int prefer_reference_element)
+void
+destroy_index_entries_sort_strings (INDICES_SORT_STRINGS *indices_sort_strings)
{
- INDEX_SORT_STRING_KEY *sort_string_key = (INDEX_SORT_STRING_KEY *)
- malloc (sizeof (INDEX_SORT_STRING_KEY));
- sort_string_key->sort_string = index_entry_element_sort_string (main_entry,
- index_entry_element, options, in_code,
- prefer_reference_element);
- set_sort_key (collation_locale, sort_string_key->sort_string,
- &sort_string_key->sort_key);
-
- return sort_string_key;
+ if (indices_sort_strings && indices_sort_strings->number)
+ {
+ size_t i;
+ for (i = 0; i < indices_sort_strings->number; i++)
+ {
+ INDEX_SORT_STRINGS *index_sort_strings
+ = &indices_sort_strings->indices[i];
+ if (index_sort_strings->entries_number > 0)
+ {
+ size_t j;
+ for (j = 0; j < index_sort_strings->entries_number; j++)
+ {
+ INDEX_ENTRY_SORT_STRING *entry_sort_string
+ = &index_sort_strings->sort_string_entries[j];
+ if (entry_sort_string->subentries_number > 0)
+ {
+ size_t k;
+ for (k = 0; k < entry_sort_string->subentries_number;
+ k++)
+ {
+ free (entry_sort_string->sort_string_subentries[k]
+ .sort_string);
+ }
+ free (entry_sort_string->sort_string_subentries);
+ }
+ }
+ }
+ free (index_sort_strings->sort_string_entries);
+ }
+ free (indices_sort_strings->indices);
+ }
+ free (indices_sort_strings);
}
-INDICES_SORTABLE_ENTRIES *
-setup_sortable_index_entries (ERROR_MESSAGE_LIST *error_messages,
- OPTIONS *options, const MERGED_INDICES *merged_indices,
- INDEX **indices_information,
- locale_t *collation_locale)
+INDICES_SORT_STRINGS *
+setup_index_entries_sort_strings (ERROR_MESSAGE_LIST *error_messages,
+ OPTIONS *options, const MERGED_INDICES *merged_indices,
+ INDEX **indices_information, int prefer_reference_element)
{
size_t i;
+ TEXT_OPTIONS *convert_text_options;
+
+ if (merged_indices->number <= 0)
+ return 0;
/* convert index entries to sort string using unicode when possible
independently of input and output encodings */
- TEXT_OPTIONS *convert_text_options = new_text_options ();
+ convert_text_options = new_text_options ();
convert_text_options->encoding = strdup ("utf-8");
/* It could be possible to set INCLUDE_DIRECTORIES, but there is no
point doing so, as it is only useful for @verbatiminclude, which
@@ -317,47 +377,37 @@ setup_sortable_index_entries (ERROR_MESSAGE_LIST
*error_messages,
options->INCLUDE_DIRECTORIES.strlist);
*/
- /* similar to using Unicode::Collate in Perl */
- *collation_locale = 0;
-
- #ifdef HAVE_STRXFRM_L
- #ifdef HAVE_NEWLOCALE
- if (!options->USE_UNICODE_COLLATION.integer == 0)
- *collation_locale = newlocale (LC_COLLATE_MASK, "en_US.utf-8", 0);
- #endif
- #endif
-
- if (merged_indices->number <= 0)
- return 0;
-
- INDICES_SORTABLE_ENTRIES *indices_sortable_entries
- = (INDICES_SORTABLE_ENTRIES *) malloc (sizeof (INDICES_SORTABLE_ENTRIES));
+ INDICES_SORT_STRINGS *indices_sort_strings
+ = (INDICES_SORT_STRINGS *) malloc (sizeof (INDICES_SORT_STRINGS));
- indices_sortable_entries->number = merged_indices->number;
- indices_sortable_entries->indices = (INDEX_SORTABLE_ENTRIES *)
- malloc (merged_indices->number * sizeof (INDEX_SORTABLE_ENTRIES));
- memset (indices_sortable_entries->indices, 0,
- merged_indices->number * sizeof (INDEX_SORTABLE_ENTRIES));
+ indices_sort_strings->number = merged_indices->number;
+ indices_sort_strings->indices = (INDEX_SORT_STRINGS *)
+ malloc (merged_indices->number * sizeof (INDEX_SORT_STRINGS));
+ memset (indices_sort_strings->indices, 0,
+ merged_indices->number * sizeof (INDEX_SORT_STRINGS));
for (i = 0; i < merged_indices->number; i++)
{
MERGED_INDEX *index = &merged_indices->indices[i];
- INDEX_SORTABLE_ENTRIES *sortable_index_entries
- = &indices_sortable_entries->indices[i];
+ INDEX_SORT_STRINGS *index_sort_strings
+ = &indices_sort_strings->indices[i];
if (index->entries_number > 0)
{
size_t j;
+ /* keep track of number of non empty index entries number */
+ size_t nr = 0;
- sortable_index_entries->index = index;
- sortable_index_entries->number = index->entries_number;
- sortable_index_entries->sortable_entries = (SORTABLE_INDEX_ENTRY *)
- malloc (index->entries_number * sizeof (SORTABLE_INDEX_ENTRY));
+ index_sort_strings->index = index;
+ index_sort_strings->sort_string_entries = (INDEX_ENTRY_SORT_STRING *)
+ malloc (index->entries_number * sizeof (INDEX_ENTRY_SORT_STRING));
for (j = 0; j < index->entries_number; j++)
{
int non_empty_index_subentries = 0;
- SORTABLE_INDEX_SUBENTRY *sortable_subentry;
+ char *sort_string;
+ INDEX_SUBENTRY_SORT_STRING *subentry_sort_string;
INDEX_ENTRY *index_entry = &index->index_entries[j];
+ INDEX_ENTRY_SORT_STRING entry_sort_string;
ELEMENT *main_entry_element = index_entry->entry_element;
ELEMENT *subentry = main_entry_element;
@@ -366,30 +416,28 @@ setup_sortable_index_entries (ERROR_MESSAGE_LIST
*error_messages,
= indices_info_index_by_name (indices_information,
index_entry->index_name);
- INDEX_SORT_STRING_KEY *sort_string_key
- = index_entry_element_sort_string_key (index_entry,
- main_entry_element, convert_text_options,
- entry_index->in_code, *collation_locale, 0);
-
- SORTABLE_INDEX_ENTRY *sortable_entry
- = &sortable_index_entries->sortable_entries[j];
+ sort_string
+ = index_entry_element_sort_string (index_entry,
+ subentry, convert_text_options,
+ entry_index->in_code,
+ prefer_reference_element);
- sortable_entry->entry = index_entry;
- sortable_entry->subentries_number = 1;
- sortable_entry->sortable_subentries = (SORTABLE_INDEX_SUBENTRY *)
- malloc (sizeof (SORTABLE_INDEX_SUBENTRY));
+ entry_sort_string.entry = index_entry;
+ entry_sort_string.subentries_number = 1;
+ entry_sort_string.sort_string_subentries
+ = (INDEX_SUBENTRY_SORT_STRING *)
+ malloc (sizeof (INDEX_SUBENTRY_SORT_STRING));
- sortable_subentry = &sortable_entry->sortable_subentries[0];
+ subentry_sort_string
+ = &entry_sort_string.sort_string_subentries[0];
- if (sort_string_key->sort_string[strspn
- (sort_string_key->sort_string, whitespace_chars)] == '\0')
+ if (sort_string[strspn
+ (sort_string, whitespace_chars)] == '\0')
{
const char *entry_cmdname;
- sortable_subentry->sort_string = strdup("");
- sortable_subentry->sort_key = strdup("");
- free (sort_string_key->sort_string);
- free (sort_string_key->sort_key);
+ subentry_sort_string->sort_string = strdup("");
+ free (sort_string);
entry_cmdname = element_command_name (main_entry_element);
if (!entry_cmdname)
@@ -404,12 +452,9 @@ setup_sortable_index_entries (ERROR_MESSAGE_LIST
*error_messages,
}
else
{
- sortable_subentry->sort_string
- = sort_string_key->sort_string;
- sortable_subentry->sort_key = sort_string_key->sort_key;
+ subentry_sort_string->sort_string = sort_string;
non_empty_index_subentries++;
}
- free (sort_string_key);
while (1)
{
@@ -419,34 +464,32 @@ setup_sortable_index_entries (ERROR_MESSAGE_LIST
*error_messages,
break;
subentry = next_subentry;
- sortable_entry->subentries_number++;
-
- sortable_entry->sortable_subentries
- = (SORTABLE_INDEX_SUBENTRY *)
- realloc (sortable_entry->sortable_subentries,
- sizeof (SORTABLE_INDEX_SUBENTRY)
- * sortable_entry->subentries_number);
- if (!sortable_entry->sortable_subentries)
+ entry_sort_string.subentries_number++;
+
+ entry_sort_string.sort_string_subentries
+ = (INDEX_SUBENTRY_SORT_STRING *)
+ realloc (entry_sort_string.sort_string_subentries,
+ sizeof (INDEX_SUBENTRY_SORT_STRING)
+ * entry_sort_string.subentries_number);
+ if (!entry_sort_string.sort_string_subentries)
fatal ("realloc failed");
- sortable_subentry
- = &sortable_entry->sortable_subentries[
- sortable_entry->subentries_number -1];
+ subentry_sort_string
+ = &entry_sort_string.sort_string_subentries[
+ entry_sort_string.subentries_number -1];
- sort_string_key
- = index_entry_element_sort_string_key (index_entry,
- subentry, convert_text_options,
- entry_index->in_code, *collation_locale, 0);
+ sort_string
+ = index_entry_element_sort_string (index_entry,
+ subentry, convert_text_options,
+ entry_index->in_code, 0);
- if (sort_string_key->sort_string[strspn
- (sort_string_key->sort_string, whitespace_chars)] == '\0')
+ if (sort_string[strspn
+ (sort_string, whitespace_chars)] == '\0')
{
const char *entry_cmdname;
- sortable_subentry->sort_string = strdup("");
- sortable_subentry->sort_key = strdup("");
- free (sort_string_key->sort_string);
- free (sort_string_key->sort_key);
+ subentry_sort_string->sort_string = strdup("");
+ free (sort_string);
entry_cmdname = element_command_name
(main_entry_element);
if (!entry_cmdname)
@@ -459,43 +502,176 @@ setup_sortable_index_entries (ERROR_MESSAGE_LIST
*error_messages,
message_list_command_warn (error_messages, options,
main_entry_element,
"empty index sub entry %zu key in @%s",
- sortable_entry->subentries_number -1,
+ entry_sort_string.subentries_number -1,
entry_cmdname);
}
else
{
- sortable_subentry->sort_string
- = sort_string_key->sort_string;
- sortable_subentry->sort_key = sort_string_key->sort_key;
+ subentry_sort_string->sort_string = sort_string;
non_empty_index_subentries++;
}
- free (sort_string_key);
}
if (non_empty_index_subentries > 0)
{
- int k;
- for (k = 0; k < sortable_entry->subentries_number; k++)
+ size_t k;
+ nr++;
+
+ for (k = 0; k < entry_sort_string.subentries_number; k++)
{
uint8_t *encoded_u8;
ucs4_t next_char;
int new_len;
- sortable_subentry
- = &sortable_entry->sortable_subentries[k];
+ subentry_sort_string
+ = &entry_sort_string.sort_string_subentries[k];
/* TODO quite inefficient, only need the first character */
encoded_u8
= u8_strconv_from_encoding (
- sortable_subentry->sort_string,
+ subentry_sort_string->sort_string,
"UTF-8", iconveh_question_mark);
new_len = u8_strmbtouc (&next_char, encoded_u8);
if (new_len > 0
&& uc_is_property (next_char,
UC_PROPERTY_ALPHABETIC))
- sortable_subentry->alpha = 1;
+ subentry_sort_string->alpha = 1;
else
- sortable_subentry->alpha = 0;
+ subentry_sort_string->alpha = 0;
free (encoded_u8);
}
+
+ memcpy (&index_sort_strings->sort_string_entries[nr-1],
+ &entry_sort_string,
+ sizeof (INDEX_ENTRY_SORT_STRING));
+ }
+ else
+ {
+ size_t k;
+ for (k = 0; k < entry_sort_string.subentries_number; k++)
+ {
+ free (entry_sort_string.sort_string_subentries[k]
+ .sort_string);
+ }
+ free (entry_sort_string.sort_string_subentries);
+ }
+ index_sort_strings->entries_number = nr;
+ }
+ }
+ }
+
+ destroy_text_options (convert_text_options);
+
+ return indices_sort_strings;
+}
+
+INDEX_COLLATOR *
+setup_collator (int use_unicode_collation, const char *collation_language,
+ const char *collation_locale)
+{
+ INDEX_COLLATOR *result = (INDEX_COLLATOR *) malloc (sizeof (INDEX_COLLATOR));
+ memset (result, 0, sizeof (INDEX_COLLATOR));
+
+ if (use_unicode_collation == 0)
+ {
+ result->type = ctn_no_unicode;
+ /* FIXME check if needed */
+ result->language = strdup ("");
+ }
+ else if (collation_language)
+ {
+ result->type = ctn_language_collation;
+ result->language = strdup (collation_language);
+ result->sv = call_setup_collator (1, collation_language);
+ }
+ else
+ {
+ #ifdef HAVE_STRXFRM_L
+ #ifdef HAVE_NEWLOCALE
+ if (collation_locale)
+ {
+ result->locale
+ = newlocale (LC_COLLATE_MASK, collation_locale, 0);
+ /* TODO warn if locale is not found? */
+ if (result->locale)
+ {
+ result->type = ctn_locale_collation;
+ result->language = strdup (collation_locale);
+ return result;
+ }
+ }
+ #endif
+ #endif
+
+ result->type = ctn_unicode;
+ /* FIXME check if needed */
+ result->language = strdup ("-");
+ result->sv = call_setup_collator (1, 0);
+ }
+ return result;
+}
+
+INDICES_SORTABLE_ENTRIES *
+setup_sortable_index_entries (INDEX_COLLATOR *collator,
+ const INDICES_SORT_STRINGS *indices_sort_strings)
+{
+ size_t i;
+
+ if (!indices_sort_strings || indices_sort_strings->number <= 0)
+ return 0;
+
+ INDICES_SORTABLE_ENTRIES *indices_sortable_entries
+ = (INDICES_SORTABLE_ENTRIES *) malloc (sizeof (INDICES_SORTABLE_ENTRIES));
+
+ indices_sortable_entries->number = indices_sort_strings->number;
+ indices_sortable_entries->indices = (INDEX_SORTABLE_ENTRIES *)
+ malloc (indices_sort_strings->number * sizeof (INDEX_SORTABLE_ENTRIES));
+ memset (indices_sortable_entries->indices, 0,
+ indices_sort_strings->number * sizeof (INDEX_SORTABLE_ENTRIES));
+
+ for (i = 0; i < indices_sort_strings->number; i++)
+ {
+ INDEX_SORT_STRINGS *index_sort_strings
+ = &indices_sort_strings->indices[i];
+ INDEX_SORTABLE_ENTRIES *sortable_index_entries
+ = &indices_sortable_entries->indices[i];
+ if (index_sort_strings->entries_number > 0)
+ {
+ size_t j;
+
+ sortable_index_entries->index = index_sort_strings->index;
+ sortable_index_entries->number = index_sort_strings->entries_number;
+ sortable_index_entries->sortable_entries = (SORTABLE_INDEX_ENTRY *)
+ malloc (index_sort_strings->entries_number
+ * sizeof (SORTABLE_INDEX_ENTRY));
+
+ for (j = 0; j < index_sort_strings->entries_number; j++)
+ {
+ size_t k;
+ INDEX_ENTRY_SORT_STRING *index_entry_sort_string
+ = &index_sort_strings->sort_string_entries[j];
+
+ SORTABLE_INDEX_ENTRY *sortable_entry
+ = &sortable_index_entries->sortable_entries[j];
+
+ sortable_entry->entry = index_entry_sort_string->entry;
+ sortable_entry->subentries_number
+ = index_entry_sort_string->subentries_number;
+ sortable_entry->sortable_subentries = (SORTABLE_INDEX_SUBENTRY *)
+ malloc (index_entry_sort_string->subentries_number
+ * sizeof (SORTABLE_INDEX_SUBENTRY));
+
+ for (k = 0; k < index_entry_sort_string->subentries_number; k++)
+ {
+ SORTABLE_INDEX_SUBENTRY *sortable_subentry
+ = &sortable_entry->sortable_subentries[k];
+ INDEX_SUBENTRY_SORT_STRING *subenty_sort_string
+ = &index_entry_sort_string->sort_string_subentries[k];
+ /* if (subenty_sort_string->sort_string) */
+ sortable_subentry->sort_string
+ /* FIXME or refer to subenty_sort_string structure? */
+ = strdup (subenty_sort_string->sort_string);
+ sortable_subentry->alpha = subenty_sort_string->alpha;
+ sortable_subentry->sort_key = get_sort_key (collator,
+ subenty_sort_string->sort_string);
}
}
}
@@ -503,9 +679,44 @@ setup_sortable_index_entries (ERROR_MESSAGE_LIST
*error_messages,
return indices_sortable_entries;
}
+
+static INDICES_SORTABLE_ENTRIES *
+setup_sort_sortable_strings_collator (
+ ERROR_MESSAGE_LIST *error_messages,
+ OPTIONS *options, int use_unicode_collation,
+ const char *collation_language,
+ const char *collation_locale,
+ const MERGED_INDICES *merged_indices,
+ INDEX **indices_information, DOCUMENT *document,
+ INDEX_COLLATOR **collator)
+{
+ const INDICES_SORT_STRINGS *indices_sort_strings;
+ INDICES_SORTABLE_ENTRIES *index_sortable_index_entries;
+
+ if (document)
+ {
+ indices_sort_strings = document_indices_sort_strings (error_messages,
+ options, document, 0);
+ }
+ else
+ {
+ indices_sort_strings = setup_index_entries_sort_strings (error_messages,
+ options, merged_indices,
+ indices_information, 0);
+ }
+
+ *collator = setup_collator (use_unicode_collation, collation_language,
+ collation_locale);
+
+ index_sortable_index_entries = setup_sortable_index_entries (*collator,
+ indices_sort_strings);
+
+ return index_sortable_index_entries;
+}
+
typedef struct LETTER_SORTABLE_ENTRIES {
char *letter;
- char *letter_sort_key;
+ BYTES_STRING *letter_sort_key;
size_t space;
size_t number;
SORTABLE_INDEX_ENTRY **sortable_entries;
@@ -517,13 +728,26 @@ typedef struct INDEX_LETTERS_SORTABLE_ENTRIES {
LETTER_SORTABLE_ENTRIES *letter_entries;
} INDEX_LETTERS_SORTABLE_ENTRIES;
+#define MIN_INT(a,b) (((a)<(b))?(a):(b))
+static int
+compare_byte_strings (BYTES_STRING *a, BYTES_STRING *b)
+{
+ size_t common_len = MIN_INT (a->len, b->len);
+ int res = memcmp (a->bytes, b->bytes, common_len);
+ if (res == 0)
+ return (a->len > b->len) - (a->len < b->len);
+ return res;
+}
+
+#undef MIN_INT
+
static int
compare_index_letter (const void *a, const void *b)
{
const LETTER_SORTABLE_ENTRIES *lse_a = (const LETTER_SORTABLE_ENTRIES *) a;
const LETTER_SORTABLE_ENTRIES *lse_b = (const LETTER_SORTABLE_ENTRIES *) b;
- return strcmp (lse_a->letter_sort_key, lse_b->letter_sort_key);
+ return compare_byte_strings (lse_a->letter_sort_key, lse_b->letter_sort_key);
}
static int
@@ -533,7 +757,7 @@ compare_sortable_subentry_keys (const void *a, const void
*b)
const SORTABLE_INDEX_SUBENTRY *sis_b = (const SORTABLE_INDEX_SUBENTRY *) b;
if (sis_a->alpha == sis_b->alpha)
- return strcmp (sis_a->sort_key, sis_b->sort_key);
+ return compare_byte_strings (sis_a->sort_key, sis_b->sort_key);
return (sis_a->alpha > sis_b->alpha) - (sis_a->alpha < sis_b->alpha);
}
@@ -581,22 +805,80 @@ compare_sortable_index_entry (const void *a, const void
*b)
return strcmp (sie_a->entry->index_name, sie_b->entry->index_name);
}
+void
+destroy_indices_sortable_entries (
+ INDICES_SORTABLE_ENTRIES *indices_sortable_entries)
+{
+ if (indices_sortable_entries)
+ {
+ size_t i;
+ for (i = 0; i < indices_sortable_entries->number; i++)
+ {
+ INDEX_SORTABLE_ENTRIES *sortable_index_entries
+ = &indices_sortable_entries->indices[i];
+ if (sortable_index_entries->number > 0)
+ {
+ size_t j;
+
+ for (j = 0; j < sortable_index_entries->number; j++)
+ {
+ size_t k;
+ SORTABLE_INDEX_ENTRY *sortable_entry
+ = &sortable_index_entries->sortable_entries[j];
+
+ for (k = 0; k < sortable_entry->subentries_number;
+ k++)
+ {
+ SORTABLE_INDEX_SUBENTRY *sortable_subentry
+ = &sortable_entry->sortable_subentries[k];
+ free (sortable_subentry->sort_key->bytes);
+ free (sortable_subentry->sort_key);
+ free (sortable_subentry->sort_string);
+ }
+ free (sortable_entry->sortable_subentries);
+ }
+ free (sortable_index_entries->sortable_entries);
+ }
+ }
+ free (indices_sortable_entries->indices);
+ free (indices_sortable_entries);
+ }
+}
+
+void
+destroy_collator (INDEX_COLLATOR *collator)
+{
+ if (collator)
+ free (collator->language);
+ free (collator);
+}
+
INDEX_SORTED_BY_LETTER *
sort_indices_by_letter (ERROR_MESSAGE_LIST *error_messages,
- OPTIONS *options, const MERGED_INDICES *merged_indices,
- INDEX **indices_information)
+ OPTIONS *options, int use_unicode_collation,
+ const char *collation_language,
+ const char *collation_locale,
+ const MERGED_INDICES *merged_indices,
+ INDEX **indices_information, DOCUMENT *document)
{
size_t i;
int index_nr = 0;
- locale_t collation_locale = 0;
static INDEX_LETTERS_SORTABLE_ENTRIES index_letters_sortable_entries;
+ INDEX_COLLATOR *collator;
INDICES_SORTABLE_ENTRIES *indices_sortable_entries
- = setup_sortable_index_entries (error_messages, options, merged_indices,
- indices_information, &collation_locale);
+ = setup_sort_sortable_strings_collator (error_messages, options,
+ use_unicode_collation, collation_language,
+ collation_locale,
+ merged_indices, indices_information,
+ document, &collator);
if (!indices_sortable_entries || indices_sortable_entries->number <= 0)
- return 0;
+ {
+ destroy_indices_sortable_entries (indices_sortable_entries);
+ destroy_collator (collator);
+ return 0;
+ }
INDEX_SORTED_BY_LETTER *sorted_index_entries
= (INDEX_SORTED_BY_LETTER *) malloc
@@ -629,7 +911,7 @@ sort_indices_by_letter (ERROR_MESSAGE_LIST *error_messages,
char *upper_letter_string;
char *norm_letter_string;
TEXT letter_text;
- char *letter_sort_key;
+ BYTES_STRING *letter_sort_key;
int letter_added = 0;
LETTER_SORTABLE_ENTRIES *letter_sortable_entries = 0;
@@ -661,6 +943,7 @@ sort_indices_by_letter (ERROR_MESSAGE_LIST *error_messages,
free (upper_letter_string);
encoded_u8 = u8_strconv_from_encoding (norm_letter_string, "UTF-8",
iconveh_question_mark);
+ free (norm_letter_string);
current_u8 = encoded_u8;
text_init (&letter_text);
@@ -692,8 +975,7 @@ sort_indices_by_letter (ERROR_MESSAGE_LIST *error_messages,
}
free (encoded_u8);
- set_sort_key (collation_locale, letter_text.text,
- &letter_sort_key);
+ letter_sort_key = get_sort_key (collator, letter_text.text);
if (index_letters_sortable_entries.letter_number > 0)
{
@@ -739,6 +1021,7 @@ sort_indices_by_letter (ERROR_MESSAGE_LIST *error_messages,
else
{
free (letter_text.text);
+ free (letter_sort_key->bytes);
free (letter_sort_key);
}
@@ -808,6 +1091,7 @@ sort_indices_by_letter (ERROR_MESSAGE_LIST *error_messages,
}
/* TODO should we reuse the letters instead? */
+ free (letter_sortable_entries->letter_sort_key->bytes);
free (letter_sortable_entries->letter_sort_key);
free (letter_sortable_entries->sortable_entries);
}
@@ -822,6 +1106,9 @@ sort_indices_by_letter (ERROR_MESSAGE_LIST *error_messages,
sorted_index_entries = realloc (sorted_index_entries,
(index_nr+1) * sizeof (INDEX_SORTED_BY_LETTER));
+ destroy_collator (collator);
+ destroy_indices_sortable_entries (indices_sortable_entries);
+
return sorted_index_entries;
}
diff --git a/tp/Texinfo/XS/main/manipulate_indices.h
b/tp/Texinfo/XS/main/manipulate_indices.h
index e558a9040a..80d748c7f7 100644
--- a/tp/Texinfo/XS/main/manipulate_indices.h
+++ b/tp/Texinfo/XS/main/manipulate_indices.h
@@ -8,7 +8,7 @@
typedef struct SORTABLE_INDEX_SUBENTRY {
char *sort_string;
- char *sort_key;
+ BYTES_STRING *sort_key;
int alpha;
} SORTABLE_INDEX_SUBENTRY;
@@ -50,11 +50,20 @@ char *index_entry_element_sort_string (const INDEX_ENTRY
*main_entry,
const ELEMENT *index_entry_element,
TEXT_OPTIONS *options, int in_code,
int prefer_reference_element);
+void destroy_index_entries_sort_strings (
+ INDICES_SORT_STRINGS *indices_sort_strings);
+INDICES_SORT_STRINGS *setup_index_entries_sort_strings (
+ ERROR_MESSAGE_LIST *error_messages,
+ OPTIONS *options, const MERGED_INDICES *merged_indices,
+ INDEX **indices_information, int prefer_reference_element);
INDEX_SORTED_BY_LETTER *sort_indices_by_letter (
ERROR_MESSAGE_LIST *error_messages,
- OPTIONS *options, const MERGED_INDICES *merged_indices,
- INDEX **indices_information);
+ OPTIONS *options, int use_unicode_collation,
+ const char *collation_language,
+ const char *collation_locale,
+ const MERGED_INDICES *merged_indices,
+ INDEX **indices_information, DOCUMENT *document);
INDEX_ENTRY_TEXT_OR_COMMAND *index_entry_first_letter_text_or_command
(INDEX_ENTRY *index_entry);
diff --git a/tp/Texinfo/options_data.txt b/tp/Texinfo/options_data.txt
index 35098056a1..64751aac9e 100644
--- a/tp/Texinfo/options_data.txt
+++ b/tp/Texinfo/options_data.txt
@@ -223,6 +223,7 @@ DEFAULT_RULE converter_customization
undef char
DEF_TABLE converter_customization undef integer
DO_ABOUT converter_customization undef integer
DOCTYPE converter_customization undef char
+DOCUMENTLANGUAGE_COLLATION converter_customization undef integer
# for LaTeX
END_USEPACKAGE converter_customization undef char
# for ext/epub3.pm
@@ -328,6 +329,9 @@ XREF_USE_NODE_NAME_ARG converter_customization
undef integer
XS_EXTERNAL_CONVERSION converter_customization undef integer
XS_EXTERNAL_FORMATTING converter_customization undef integer
+# used for development tests, not sure that it will stay in the long term
+XS_STRXFRM_COLLATION_LOCALE converter_customization undef char
+
# Not strings
LINKS_BUTTONS converter_other undef buttons
TOP_BUTTONS converter_other undef buttons
diff --git a/tp/t/test_utils.pl b/tp/t/test_utils.pl
index 5d828b39ae..fb1553fc6d 100644
--- a/tp/t/test_utils.pl
+++ b/tp/t/test_utils.pl
@@ -1147,6 +1147,10 @@ sub test($$)
}
}
}
+
+ Texinfo::Document::indices_sort_strings($registrar,
+ $main_configuration, $document);
+
# could be in a if !$XS_structuring, but the function should not be
# overriden already in that case
$document = Texinfo::Document::rebuild_document($document);
@@ -1162,7 +1166,6 @@ sub test($$)
}
my ($errors, $error_nrs) = $registrar->errors();
-
my $indices_information = $document->indices_information();
# FIXME maybe it would be good to compare $merged_index_entries?
my $merged_index_entries = $document->merged_indices();
@@ -1190,6 +1193,7 @@ sub test($$)
my $indices_sort_strings
= Texinfo::Document::indices_sort_strings($registrar,
$main_configuration, $document);
+
$index_entries_sort_strings
= Texinfo::Indices::format_index_entries_sort_strings(
$indices_sort_strings);
@@ -1213,6 +1217,7 @@ sub test($$)
}
}
+
# use the parser expanded formats to be similar to the main program,
# and also to avoid having @inline* and raw output format @-commands
# with elided contents especially parsed because they are ignored
diff --git a/tp/tests/coverage/res_parser/formatting_epub/formatting.2
b/tp/tests/coverage/res_parser/formatting_epub/formatting.2
index 2d0c9ff589..781e7d8016 100644
--- a/tp/tests/coverage/res_parser/formatting_epub/formatting.2
+++ b/tp/tests/coverage/res_parser/formatting_epub/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:32: warning: no htmlxref.cnf entry found for `file n---ame@'
(possibly involving @mymacro)
formatting.texi:32: warning: no htmlxref.cnf entry found for `file name'
(possibly involving @mymacro)
formatting.texi:32: warning: no htmlxref.cnf entry found for `a comma, in
file' (possibly involving @mymacro)
@@ -371,4 +372,3 @@ formatting.texi:88: warning: @image file `f--ile@.' (for
HTML) not found, using
texi2any: @image file `f--ile@.' can not be copied
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
texi2any: @image file `filejk _" %@' can not be copied
-formatting.texi:123: warning: empty index key in @findex
diff --git a/tp/tests/coverage/res_parser/formatting_html32/formatting.2
b/tp/tests/coverage/res_parser/formatting_html32/formatting.2
index c948381bb4..c4a9decc65 100644
--- a/tp/tests/coverage/res_parser/formatting_html32/formatting.2
+++ b/tp/tests/coverage/res_parser/formatting_html32/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
@@ -327,4 +328,3 @@ formatting.texi:88: warning: @image file `f-ile' (for HTML)
not found, using `f-
formatting.texi:88: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git a/tp/tests/coverage/res_parser/formatting_html_no_split/formatting.2
b/tp/tests/coverage/res_parser/formatting_html_no_split/formatting.2
index 082b249eb9..e84c66d60d 100644
--- a/tp/tests/coverage/res_parser/formatting_html_no_split/formatting.2
+++ b/tp/tests/coverage/res_parser/formatting_html_no_split/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:22: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:22: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:22: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
@@ -319,4 +320,3 @@ formatting.texi:88: warning: @image file `f-ile' (for HTML)
not found, using `f-
formatting.texi:88: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git a/tp/tests/coverage/res_parser/formatting_info/formatting.2
b/tp/tests/coverage/res_parser/formatting_info/formatting.2
index e8ca7e49e9..5047e36266 100644
--- a/tp/tests/coverage/res_parser/formatting_info/formatting.2
+++ b/tp/tests/coverage/res_parser/formatting_info/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:22: warning: @ref node name should not contain `,' (possibly
involving @mymacro)
formatting.texi:22: warning: @inforef node name should not contain `,'
(possibly involving @mymacro)
formatting.texi:22: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
@@ -323,5 +324,4 @@ formatting.texi:88: warning: @ref node name should not
contain `,' (possibly inv
formatting.texi:88: warning: @inforef node name should not contain `,'
(possibly involving @mymacro)
formatting.texi:88: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
formatting.texi:88: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
formatting.texi:194: warning: @node name should not contain `,': s--ect,ion
diff --git a/tp/tests/coverage/res_parser/formatting_plaintext/formatting.2
b/tp/tests/coverage/res_parser/formatting_plaintext/formatting.2
index 9db7ff0fc3..57582a5f16 100644
--- a/tp/tests/coverage/res_parser/formatting_plaintext/formatting.2
+++ b/tp/tests/coverage/res_parser/formatting_plaintext/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:22: warning: @ref node name should not contain `,' (possibly
involving @mymacro)
formatting.texi:22: warning: @inforef node name should not contain `,'
(possibly involving @mymacro)
formatting.texi:22: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
@@ -319,4 +320,3 @@ formatting.texi:88: warning: @ref node name should not
contain `,' (possibly inv
formatting.texi:88: warning: @inforef node name should not contain `,'
(possibly involving @mymacro)
formatting.texi:88: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
formatting.texi:88: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git a/tp/tests/coverage/res_parser/formatting_xhtml/formatting.2
b/tp/tests/coverage/res_parser/formatting_xhtml/formatting.2
index c948381bb4..c4a9decc65 100644
--- a/tp/tests/coverage/res_parser/formatting_xhtml/formatting.2
+++ b/tp/tests/coverage/res_parser/formatting_xhtml/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
@@ -327,4 +328,3 @@ formatting.texi:88: warning: @image file `f-ile' (for HTML)
not found, using `f-
formatting.texi:88: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git a/tp/tests/layout/res_parser/formatting_enable_encoding/formatting.2
b/tp/tests/layout/res_parser/formatting_enable_encoding/formatting.2
index c948381bb4..c4a9decc65 100644
--- a/tp/tests/layout/res_parser/formatting_enable_encoding/formatting.2
+++ b/tp/tests/layout/res_parser/formatting_enable_encoding/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
@@ -327,4 +328,3 @@ formatting.texi:88: warning: @image file `f-ile' (for HTML)
not found, using `f-
formatting.texi:88: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git a/tp/tests/layout/res_parser/formatting_epub_nodes/formatting.2
b/tp/tests/layout/res_parser/formatting_epub_nodes/formatting.2
index 2d0c9ff589..781e7d8016 100644
--- a/tp/tests/layout/res_parser/formatting_epub_nodes/formatting.2
+++ b/tp/tests/layout/res_parser/formatting_epub_nodes/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:32: warning: no htmlxref.cnf entry found for `file n---ame@'
(possibly involving @mymacro)
formatting.texi:32: warning: no htmlxref.cnf entry found for `file name'
(possibly involving @mymacro)
formatting.texi:32: warning: no htmlxref.cnf entry found for `a comma, in
file' (possibly involving @mymacro)
@@ -371,4 +372,3 @@ formatting.texi:88: warning: @image file `f--ile@.' (for
HTML) not found, using
texi2any: @image file `f--ile@.' can not be copied
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
texi2any: @image file `filejk _" %@' can not be copied
-formatting.texi:123: warning: empty index key in @findex
diff --git a/tp/tests/layout/res_parser/formatting_exotic/formatting.2
b/tp/tests/layout/res_parser/formatting_exotic/formatting.2
index 082b249eb9..e84c66d60d 100644
--- a/tp/tests/layout/res_parser/formatting_exotic/formatting.2
+++ b/tp/tests/layout/res_parser/formatting_exotic/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:22: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:22: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:22: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
@@ -319,4 +320,3 @@ formatting.texi:88: warning: @image file `f-ile' (for HTML)
not found, using `f-
formatting.texi:88: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git a/tp/tests/layout/res_parser/formatting_fr/formatting.2
b/tp/tests/layout/res_parser/formatting_fr/formatting.2
index c948381bb4..c4a9decc65 100644
--- a/tp/tests/layout/res_parser/formatting_fr/formatting.2
+++ b/tp/tests/layout/res_parser/formatting_fr/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
@@ -327,4 +328,3 @@ formatting.texi:88: warning: @image file `f-ile' (for HTML)
not found, using `f-
formatting.texi:88: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git a/tp/tests/layout/res_parser/formatting_fr_icons/formatting.2
b/tp/tests/layout/res_parser/formatting_fr_icons/formatting.2
index c948381bb4..c4a9decc65 100644
--- a/tp/tests/layout/res_parser/formatting_fr_icons/formatting.2
+++ b/tp/tests/layout/res_parser/formatting_fr_icons/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
@@ -327,4 +328,3 @@ formatting.texi:88: warning: @image file `f-ile' (for HTML)
not found, using `f-
formatting.texi:88: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git a/tp/tests/layout/res_parser/formatting_fr_info/formatting.2
b/tp/tests/layout/res_parser/formatting_fr_info/formatting.2
index e8ca7e49e9..5047e36266 100644
--- a/tp/tests/layout/res_parser/formatting_fr_info/formatting.2
+++ b/tp/tests/layout/res_parser/formatting_fr_info/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:22: warning: @ref node name should not contain `,' (possibly
involving @mymacro)
formatting.texi:22: warning: @inforef node name should not contain `,'
(possibly involving @mymacro)
formatting.texi:22: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
@@ -323,5 +324,4 @@ formatting.texi:88: warning: @ref node name should not
contain `,' (possibly inv
formatting.texi:88: warning: @inforef node name should not contain `,'
(possibly involving @mymacro)
formatting.texi:88: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
formatting.texi:88: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
formatting.texi:194: warning: @node name should not contain `,': s--ect,ion
diff --git
a/tp/tests/layout/res_parser/formatting_info_ascii_punctuation/formatting.2
b/tp/tests/layout/res_parser/formatting_info_ascii_punctuation/formatting.2
index e8ca7e49e9..5047e36266 100644
--- a/tp/tests/layout/res_parser/formatting_info_ascii_punctuation/formatting.2
+++ b/tp/tests/layout/res_parser/formatting_info_ascii_punctuation/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:22: warning: @ref node name should not contain `,' (possibly
involving @mymacro)
formatting.texi:22: warning: @inforef node name should not contain `,'
(possibly involving @mymacro)
formatting.texi:22: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
@@ -323,5 +324,4 @@ formatting.texi:88: warning: @ref node name should not
contain `,' (possibly inv
formatting.texi:88: warning: @inforef node name should not contain `,'
(possibly involving @mymacro)
formatting.texi:88: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
formatting.texi:88: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
formatting.texi:194: warning: @node name should not contain `,': s--ect,ion
diff --git
a/tp/tests/layout/res_parser/formatting_info_disable_encoding/formatting.2
b/tp/tests/layout/res_parser/formatting_info_disable_encoding/formatting.2
index e8ca7e49e9..5047e36266 100644
--- a/tp/tests/layout/res_parser/formatting_info_disable_encoding/formatting.2
+++ b/tp/tests/layout/res_parser/formatting_info_disable_encoding/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:22: warning: @ref node name should not contain `,' (possibly
involving @mymacro)
formatting.texi:22: warning: @inforef node name should not contain `,'
(possibly involving @mymacro)
formatting.texi:22: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
@@ -323,5 +324,4 @@ formatting.texi:88: warning: @ref node name should not
contain `,' (possibly inv
formatting.texi:88: warning: @inforef node name should not contain `,'
(possibly involving @mymacro)
formatting.texi:88: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
formatting.texi:88: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
formatting.texi:194: warning: @node name should not contain `,': s--ect,ion
diff --git a/tp/tests/layout/res_parser/formatting_inline_css/formatting.2
b/tp/tests/layout/res_parser/formatting_inline_css/formatting.2
index c948381bb4..c4a9decc65 100644
--- a/tp/tests/layout/res_parser/formatting_inline_css/formatting.2
+++ b/tp/tests/layout/res_parser/formatting_inline_css/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
@@ -327,4 +328,3 @@ formatting.texi:88: warning: @image file `f-ile' (for HTML)
not found, using `f-
formatting.texi:88: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git a/tp/tests/layout/res_parser/formatting_mathjax/formatting.2
b/tp/tests/layout/res_parser/formatting_mathjax/formatting.2
index c948381bb4..c4a9decc65 100644
--- a/tp/tests/layout/res_parser/formatting_mathjax/formatting.2
+++ b/tp/tests/layout/res_parser/formatting_mathjax/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
@@ -327,4 +328,3 @@ formatting.texi:88: warning: @image file `f-ile' (for HTML)
not found, using `f-
formatting.texi:88: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git
a/tp/tests/layout/res_parser/formatting_numerical_entities/formatting.2
b/tp/tests/layout/res_parser/formatting_numerical_entities/formatting.2
index c948381bb4..c4a9decc65 100644
--- a/tp/tests/layout/res_parser/formatting_numerical_entities/formatting.2
+++ b/tp/tests/layout/res_parser/formatting_numerical_entities/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
@@ -327,4 +328,3 @@ formatting.texi:88: warning: @image file `f-ile' (for HTML)
not found, using `f-
formatting.texi:88: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git
a/tp/tests/layout/res_parser/formatting_plaintext_ascii_punctuation/formatting.2
b/tp/tests/layout/res_parser/formatting_plaintext_ascii_punctuation/formatting.2
index 9db7ff0fc3..57582a5f16 100644
---
a/tp/tests/layout/res_parser/formatting_plaintext_ascii_punctuation/formatting.2
+++
b/tp/tests/layout/res_parser/formatting_plaintext_ascii_punctuation/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:22: warning: @ref node name should not contain `,' (possibly
involving @mymacro)
formatting.texi:22: warning: @inforef node name should not contain `,'
(possibly involving @mymacro)
formatting.texi:22: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
@@ -319,4 +320,3 @@ formatting.texi:88: warning: @ref node name should not
contain `,' (possibly inv
formatting.texi:88: warning: @inforef node name should not contain `,'
(possibly involving @mymacro)
formatting.texi:88: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
formatting.texi:88: warning: @ref cross-reference name should not contain `:'
(possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git
a/tp/tests/layout/res_parser/formatting_sort_element_counts/formatting.2
b/tp/tests/layout/res_parser/formatting_sort_element_counts/formatting.2
index 082b249eb9..e84c66d60d 100644
--- a/tp/tests/layout/res_parser/formatting_sort_element_counts/formatting.2
+++ b/tp/tests/layout/res_parser/formatting_sort_element_counts/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:22: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:22: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:22: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
@@ -319,4 +320,3 @@ formatting.texi:88: warning: @image file `f-ile' (for HTML)
not found, using `f-
formatting.texi:88: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git a/tp/tests/layout/res_parser/formatting_texi2html/formatting.2
b/tp/tests/layout/res_parser/formatting_texi2html/formatting.2
index c948381bb4..c4a9decc65 100644
--- a/tp/tests/layout/res_parser/formatting_texi2html/formatting.2
+++ b/tp/tests/layout/res_parser/formatting_texi2html/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
@@ -327,4 +328,3 @@ formatting.texi:88: warning: @image file `f-ile' (for HTML)
not found, using `f-
formatting.texi:88: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git a/tp/tests/layout/res_parser/formatting_texi2html_nodes/formatting.2
b/tp/tests/layout/res_parser/formatting_texi2html_nodes/formatting.2
index c948381bb4..c4a9decc65 100644
--- a/tp/tests/layout/res_parser/formatting_texi2html_nodes/formatting.2
+++ b/tp/tests/layout/res_parser/formatting_texi2html_nodes/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
@@ -327,4 +328,3 @@ formatting.texi:88: warning: @image file `f-ile' (for HTML)
not found, using `f-
formatting.texi:88: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git a/tp/tests/layout/res_parser/formatting_weird_quotes/formatting.2
b/tp/tests/layout/res_parser/formatting_weird_quotes/formatting.2
index c948381bb4..c4a9decc65 100644
--- a/tp/tests/layout/res_parser/formatting_weird_quotes/formatting.2
+++ b/tp/tests/layout/res_parser/formatting_weird_quotes/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:32: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
@@ -327,4 +328,3 @@ formatting.texi:88: warning: @image file `f-ile' (for HTML)
not found, using `f-
formatting.texi:88: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git a/tp/tests/tex_html/res_parser/formatting_singular/formatting.2
b/tp/tests/tex_html/res_parser/formatting_singular/formatting.2
index 004f7d9618..3aa1d61f84 100644
--- a/tp/tests/tex_html/res_parser/formatting_singular/formatting.2
+++ b/tp/tests/tex_html/res_parser/formatting_singular/formatting.2
@@ -307,6 +307,7 @@ formatting.texi:88: @ref reference to nonexistent node
`node' (possibly involvin
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
formatting.texi:88: @ref reference to nonexistent node `node' (possibly
involving @mymacro)
+formatting.texi:123: warning: empty index key in @findex
texexpand
formatting.texi:22: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:22: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
@@ -320,4 +321,3 @@ formatting.texi:88: warning: @image file `f-ile' (for HTML)
not found, using `f-
formatting.texi:88: warning: @image file `f-ile' (for HTML) not found, using
`f-ile.jpg' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `f--ile@.' (for HTML) not found,
using `f--ile@..file ext e--xt}' (possibly involving @mymacro)
formatting.texi:88: warning: @image file `filejk _" %@' (for HTML) not found,
using `filejk _" %@.jpg' (possibly involving @mymacro)
-formatting.texi:123: warning: empty index key in @findex
diff --git a/tp/texi2any.pl b/tp/texi2any.pl
index e7aeeb452b..8577b6a008 100755
--- a/tp/texi2any.pl
+++ b/tp/texi2any.pl
@@ -617,12 +617,14 @@ my %formats_table = (
'info' => {
'nodes_tree' => 1,
'floats' => 1,
+ 'setup_index_entries_sort_strings' => 1,
'module' => 'Texinfo::Convert::Info'
},
'plaintext' => {
'nodes_tree' => 1,
'floats' => 1,
'split' => 1,
+ 'setup_index_entries_sort_strings' => 1,
'module' => 'Texinfo::Convert::Plaintext'
},
'html' => {
@@ -633,6 +635,7 @@ my %formats_table = (
'move_index_entries_after_items' => 1,
'relate_index_entries_to_table_items' => 1,
'no_warn_non_empty_parts' => 1,
+ 'setup_index_entries_sort_strings' => 1,
'module' => 'Texinfo::Convert::HTML'
},
'latex' => {
@@ -657,6 +660,7 @@ my %formats_table = (
# is 'ixinsxml', as Texinfo tree conversion is done
# from within Texinfo::Convert::IXINSXML
'nodes_tree' => 1,
+ 'setup_index_entries_sort_strings' => 1,
'module' => 'Texinfo::Convert::IXINSXML',
'floats' => 1,
},
@@ -1671,6 +1675,11 @@ while(@input_files) {
Texinfo::Structuring::number_floats($document);
}
+ if ($formats_table{$converted_format}->{'setup_index_entries_sort_strings'})
{
+ Texinfo::Document::indices_sort_strings($registrar, $main_configuration,
+ $document);
+ }
+
$document = Texinfo::Document::rebuild_document($document);
if ($XS_structuring) {