Attachment 120046 Details for Bug 55977 – Fix finding Arabic Presentation Forms ligatures

[patch] Fix finding Arabic Presentation Forms ligatures

0002-Fix-finding-Arabic-Presentation-Forms-ligatures.patch (text/plain), 3.90 KB, created by Khaled Hosny on 2015-11-23 10:01:58 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Khaled Hosny

Created: 2015-11-23 10:01:58 UTC

Size: 3.90 KB

patch

obsolete

>From 773a46c257626f59074abe7827c7ac2391fdabd0 Mon Sep 17 00:00:00 2001
>From: Khaled Hosny <khaledhosny@eglug.org>
>Date: Mon, 23 Nov 2015 13:52:10 +0400
>Subject: [PATCH 2/2] Fix finding Arabic Presentation Forms ligatures
>
>PDF text containing Arabic Presentation forms ligatures is still not
>found after the previous commit.
>
>This because the ligatures are decomposed in logical order after
>normalisation, while the whole string is in visual order.  For example
>the RTL text ABCD in visual order will be DCBA, and assuming B is a
>ligature, it will be decomposed to B1B2 so the string after
>normalization will be DCB1B2A while we are expecting it to be DCB2B1A.
>
>This patch reverses the order of the decomposition of RTL characters to
>work around this issue.
>---
> poppler/TextOutputDev.cc    |  3 ++-
> poppler/UnicodeTypeTable.cc | 14 ++++++++++----
> poppler/UnicodeTypeTable.h  |  5 +++--
> 3 files changed, 15 insertions(+), 7 deletions(-)
>
>diff --git a/poppler/TextOutputDev.cc b/poppler/TextOutputDev.cc
>index 31d303d..cf74d23 100644
>--- a/poppler/TextOutputDev.cc
>+++ b/poppler/TextOutputDev.cc
>@@ -3900,7 +3900,8 @@ GBool TextPage::findText(Unicode *s, int len,
>       if (!line->normalized)
> 	line->normalized = unicodeNormalizeNFKC(line->text, line->len, 
> 						&line->normalized_len, 
>-						&line->normalized_idx);
>+						&line->normalized_idx,
>+						true);
>       // convert the line to uppercase
>       m = line->normalized_len;
>       if (!caseSensitive) {
>diff --git a/poppler/UnicodeTypeTable.cc b/poppler/UnicodeTypeTable.cc
>index 721af9d..8f574e5 100644
>--- a/poppler/UnicodeTypeTable.cc
>+++ b/poppler/UnicodeTypeTable.cc
>@@ -1015,7 +1015,9 @@ Unicode unicodeToUpper(Unicode c) {
> // of characters written. @buf may be NULL, in which case the length of the
> // decomposition is returned but nothing is written. If @u is its own
> // decomposition, write @u into @buf and return 1.
>-static int decomp_compat(Unicode u, Unicode *buf) {
>+// If reverseRTL is true, then decompositions of RTL characters will be output
>+// in reverse order.
>+static int decomp_compat(Unicode u, Unicode *buf, GBool reverseRTL = false) {
>   // decomposition tables stored as lists {character, decomp_length, offset}
>   // so we do a binary search
>   int start = 0, end = DECOMP_TABLE_LENGTH;
>@@ -1031,7 +1033,10 @@ static int decomp_compat(Unicode u, Unicode *buf) {
> 	  int length = decomp_table[midpoint].length, i;
> 	  if (buf)
> 	    for (i = 0; i < length; ++i)
>-	      buf[i] = decomp_expansion[offset + i];
>+		if (unicodeTypeR(u) && reverseRTL)
>+		  buf[i] = decomp_expansion[offset + length - i - 1];
>+		else
>+		  buf[i] = decomp_expansion[offset + i];
> 	  return length;
> 	}
>       } else if (midpoint == start)
>@@ -1126,7 +1131,8 @@ static GBool combine(Unicode base, Unicode add, Unicode *out) {
> // corresponding unnormalized character. @indices is not guaranteed monotone or
> // onto.
> Unicode *unicodeNormalizeNFKC(Unicode *in, int len, 
>-			      int *out_len, int **indices) {
>+			      int *out_len, int **indices,
>+			      GBool reverseRtlCompat) {
>   Unicode *out;
>   int i, o, *classes, *idx = NULL;
> 
>@@ -1174,7 +1180,7 @@ Unicode *unicodeNormalizeNFKC(Unicode *in, int len,
> 	u = in[j];
> 	if (j != i && COMBINING_CLASS(u) == 0)
> 	  break;
>-	dlen = decomp_compat(u, out + p);
>+	dlen = decomp_compat(u, out + p, reverseRtlCompat);
> 	for (q = p; q < p + dlen; ++q) {
> 	  classes[q] = COMBINING_CLASS(out[q]);
> 	  if (indices)
>diff --git a/poppler/UnicodeTypeTable.h b/poppler/UnicodeTypeTable.h
>index 869aad9..6d10a67 100644
>--- a/poppler/UnicodeTypeTable.h
>+++ b/poppler/UnicodeTypeTable.h
>@@ -38,7 +38,8 @@ extern GBool unicodeIsAlphabeticPresentationForm(Unicode c);
> 
> extern Unicode unicodeToUpper(Unicode c);
> 
>-extern Unicode *unicodeNormalizeNFKC(Unicode *in, int len, 
>-				     int *out_len, int **offsets);
>+extern Unicode *unicodeNormalizeNFKC(Unicode *in, int len,
>+				     int *out_len, int **offsets,
>+				     GBool reverseRtlCompat = false);
> 
> #endif
>-- 
>2.6.2
>

Actions: View | Splinter Review

Attachments on bug 55977: 68555 | 68556 | 68652 | 68860 | 69530 | 70723 | 73767 | 82433 | 88773 | 88871 | 91475 | 117108 | 117118 | 117145 | 117148 | 119906 | 119908 | 120046

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.