tFix choiceoverride - transliterate - Transliteration engine
HTML git clone git://lumidify.org/transliterate.git
DIR Log
DIR Files
DIR Refs
DIR README
DIR LICENSE
---
DIR commit 659fb098117bc50efc4cb92d586afa267c727489
DIR parent 9ceb2478425536bef95dcf3d4cc4e4c2d0bd51f4
HTML Author: lumidify <nobody@lumidify.org>
Date: Mon, 6 Apr 2020 15:43:26 +0200
Fix choiceoverride
Diffstat:
A tests/data/override.txt | 1 +
A tests/test6/README | 2 ++
A tests/test6/config | 16 ++++++++++++++++
A tests/test6/descr.txt | 1 +
A tests/test6/err.txt | 6 ++++++
A tests/test6/expected.txt | 6 ++++++
A tests/test6/input.txt | 6 ++++++
M transliterate.pl | 16 ++++++++++++----
8 files changed, 50 insertions(+), 4 deletions(-)
---
DIR diff --git a/tests/data/override.txt b/tests/data/override.txt
t@@ -0,0 +1 @@
+word0_replaced$word0_replaced2 word0_replaced2
DIR diff --git a/tests/test6/README b/tests/test6/README
t@@ -0,0 +1,2 @@
+This can be used to play around with "choiceoverride", but it won't
+work with "--nochoices".
DIR diff --git a/tests/test6/config b/tests/test6/config
t@@ -0,0 +1,16 @@
+split "[ \n]+"
+beforeword " "
+afterword "[ \n]"
+
+ignore "../data/ignore.txt"
+table words "../data/words.txt"
+table endings "../data/endings.txt"
+choiceoverride "../data/override.txt"
+
+expand words endings
+
+match "\d+" "num_replaced" beginword
+
+group beginword endword
+replace words
+endgroup
DIR diff --git a/tests/test6/descr.txt b/tests/test6/descr.txt
t@@ -0,0 +1 @@
+Basic test
DIR diff --git a/tests/test6/err.txt b/tests/test6/err.txt
t@@ -0,0 +1,6 @@
+Unknown word: "word20"
+Unknown word: "word01231"
+Word "word0_replaced$word0_replaced2" with 2 word choices.
+Unknown word: "aword1"
+Unknown word: "end3"
+Word "word0_replacedend3_replaced$word0_replaced2end3_replaced" with 2 word choices.
DIR diff --git a/tests/test6/expected.txt b/tests/test6/expected.txt
t@@ -0,0 +1,6 @@
+word1_replaced word2_replaced
+num_replacedword1_replaced word9_replaced num_replaced word4_replaced
+word20 word01231 word0_replaced2
+aword1
+word1_replacedend1_replaced word0_replacedend3_replaced$word0_replaced2end3_replaced end3
+num_replacedword2_replacedend1_replaced
DIR diff --git a/tests/test6/input.txt b/tests/test6/input.txt
t@@ -0,0 +1,6 @@
+word1 word2
+123word1 word9 123 word4
+word20 word01231 word0
+aword1
+word1end1 word0end3 end3
+432word2end1
DIR diff --git a/transliterate.pl b/transliterate.pl
t@@ -5,6 +5,9 @@
# NOTE: If you're wondering why the error codes used by the functions are so
# inconsistent, go ask my former self
+# NOTE 2: This codebase has grown as new features were needed, so it's quite
+# ugly now, but I haven't had time to clean it up.
+
use strict;
use warnings;
use utf8;
t@@ -525,7 +528,7 @@ sub open_file_rel_abs {
# Load a file of replacement words into a hash table
sub load_table {
- my ($filename, $args, $config, $revert) = @_;
+ my ($filename, $args, $config, $src_verbatim, $revert) = @_;
my $fh = open_file_rel_abs $filename, $args->{"config"};
return if !$fh;
my %table;
t@@ -547,7 +550,7 @@ sub load_table {
$word = NFD $words[0];
$replacement = NFD $words[1];
}
- my @word_choices = split /\Q$config->{choicesep}\E/, $word;
+ my @word_choices = $src_verbatim ? ($word) : split /\Q$config->{choicesep}\E/, $word;
foreach my $word_choice (@word_choices) {
if (exists $table{$word_choice}) {
if ($args->{"checkduplicates"}) {
t@@ -722,7 +725,7 @@ sub interpret_config {
if (exists $path_to_table{$table_path}) {
$table = $path_to_table{$table_path};
} else {
- $table = load_table $table_path, $args, \%config, $table_args{"revert"};
+ $table = load_table $table_path, $args, \%config, 0, $table_args{"revert"};
return if !defined $table;
$path_to_table{$table_path} = $table;
}
t@@ -737,7 +740,10 @@ sub interpret_config {
$config{"display_tables"}->{$table_path} = 1 if !exists $table_args{"nodisplay"};
} elsif ($cmd_name eq "choiceoverride") {
my $table_path = $cmd->[1]->{"value"};
- my $table = load_table $table_path, $args, \%config;
+ # argument $src_verbatim to load_table forces it to take the entire
+ # source word without splitting it up with "choicesep", since here,
+ # we explicitly want to replace multiple choices with one
+ my $table = load_table $table_path, $args, \%config, 1;
return if !defined $table;
if (exists $config{"choiceoverride"}) {
warn "Duplicate specification of \"choiceoverride\" option.\n";
t@@ -1825,6 +1831,8 @@ Note that this does not sort the choices before comparison and they have to
be matched exactly, so when a new choice is added, that needs to be added to
this mapping as well, in exactly the same order.
+Like B<targetdiacritics>, this has no effect if B<--nochoices> is set.
+
To clarify the order in which choices are added (if they are not explicitly
specified):