diff options
author | Sergey Poznyakoff <gray@gnu.org.ua> | 2015-07-14 16:09:36 +0300 |
---|---|---|
committer | Sergey Poznyakoff <gray@gnu.org.ua> | 2015-07-14 16:09:36 +0300 |
commit | 4097896542f2279700794104c7c0728beed38cd0 (patch) | |
tree | 664012755efc30126a0ccef42245ce3068f06adc | |
parent | 28072898f1bd9a925d73ac187d560198d6345524 (diff) | |
download | wikitrans-4097896542f2279700794104c7c0728beed38cd0.tar.gz wikitrans-4097896542f2279700794104c7c0728beed38cd0.tar.bz2 |
Fix processing of environments (numbered/unnumbered lists, definition lists) and indented strings.
* wikimarkup.py (tokread): Always add 'continuation' key to DELIM
entries. Delete whitespace following environment delimiters.
(peektkn): Take an optional offset argument.
(parse0): Handle indentations.
* wiki2html.py (str_ind): Use <dl> to produce indentations.
* wiki2texi.py (str_ind): End text with a newline.
* testdata/colon.html: Update.
* testdata/deflist.html: Update.
* testdata/numlist.html: Update.
* testdata/unlist.html: Update.
-rw-r--r-- | testdata/colon.html | 4 | ||||
-rw-r--r-- | testdata/deflist.html | 3 | ||||
-rw-r--r-- | testdata/numlist.html | 2 | ||||
-rw-r--r-- | testdata/unlist.html | 2 | ||||
-rw-r--r-- | wiki2html.py | 2 | ||||
-rw-r--r-- | wiki2texi.py | 2 | ||||
-rw-r--r-- | wikimarkup.py | 33 |
7 files changed, 29 insertions, 19 deletions
diff --git a/testdata/colon.html b/testdata/colon.html index b9e35ed..2fb6ed3 100644 --- a/testdata/colon.html +++ b/testdata/colon.html | |||
@@ -1,4 +1,4 @@ | |||
1 | <dl><dd> A colon (:) indents a line or paragraph.</dd></dl><p>A newline starts a new paragraph. | 1 | <dl><dd>A colon (:) indents a line or paragraph.</dd></dl><p>A newline starts a new paragraph. |
2 | Should only be used on talk pages. | 2 | Should only be used on talk pages. |
3 | For articles, you probably want the blockquote tag.</p><dl><dd> We use 1 colon to indent once.<dl><dd> We use 2 colons to indent twice.<dl><dd> 3 colons to indent 3 times, and so on.</dd></dl></dd></dl></dd></dl> | 3 | For articles, you probably want the blockquote tag.</p><dl><dd>We use 1 colon to indent once.</dd></dl><dl><dd><dl><dd>We use 2 colons to indent twice.</dd></dl></dd></dl><dl><dd><dl><dd><dl><dd>3 colons to indent 3 times, and so on.</dd></dl></dd></dl></dd></dl> |
4 | 4 | ||
diff --git a/testdata/deflist.html b/testdata/deflist.html index 8986533..774f21f 100644 --- a/testdata/deflist.html +++ b/testdata/deflist.html | |||
@@ -1,2 +1 @@ | |||
1 | <dl><dt>item 1</dt><dd> definition 1</dd><dt>item 2</dt><dd> definition 2-1</dd><dd> definition 2-2</dd></dl> | <dl><dt>item 1</dt><dd>definition 1</dd><dt>item 2</dt><dd>definition 2-1</dd><dd>definition 2-2</dd></dl> | |
2 | |||
diff --git a/testdata/numlist.html b/testdata/numlist.html index ce9dd2d..0ce8a9a 100644 --- a/testdata/numlist.html +++ b/testdata/numlist.html | |||
@@ -1,2 +1,2 @@ | |||
1 | <ol><li> <i>Numbered lists</i> are:<ol><li> Very organized</li><li> Easy to follow</li></ol></li></ol><p>A newline marks the end of the list.</p><ol><li> New numbering starts with 1.</li></ol> | 1 | <ol><li><i>Numbered lists</i> are:<ol><li>Very organized</li><li>Easy to follow</li></ol></li></ol><p>A newline marks the end of the list.</p><ol><li>New numbering starts with 1.</li></ol> |
2 | 2 | ||
diff --git a/testdata/unlist.html b/testdata/unlist.html index 87be30f..ac4d32b 100644 --- a/testdata/unlist.html +++ b/testdata/unlist.html | |||
@@ -1,2 +1,2 @@ | |||
1 | <ul><li> <i>Unordered lists</i> are easy to do:<ul><li> Start every line with a star.<ul><li> More stars indicate a deeper level.</li></ul></li></ul> Previous item continues.<ul><li> A newline</li></ul></li><li> in a list </li></ul><p>marks the end of the list.</p><ul><li>Of course you can start again.</li></ul> | 1 | <ul><li><i>Unordered lists</i> are easy to do:<ul><li>Start every line with a star.<ul><li>More stars indicate a deeper level.</li></ul>Previous item continues.</li><li>A newline</li></ul></li><li>in a list </li></ul><p>marks the end of the list.</p><ul><li>Of course you can start again.</li></ul> |
2 | 2 | ||
diff --git a/wiki2html.py b/wiki2html.py index 66939c4..f3ea0e3 100644 --- a/wiki2html.py +++ b/wiki2html.py | |||
@@ -205,3 +205,3 @@ class HtmlWikiMarkup (WikiMarkup): | |||
205 | def str_ind(self, elt): | 205 | def str_ind(self, elt): |
206 | return (" " * 2 * elt['level']) + self.format(elt['content']) | 206 | return ("<dl><dd>" * elt['level']) + self.format(elt['content']) + "</dd></dl>" * elt['level'] |
207 | 207 | ||
diff --git a/wiki2texi.py b/wiki2texi.py index 0b3eb77..e9009ec 100644 --- a/wiki2texi.py +++ b/wiki2texi.py | |||
@@ -188,3 +188,3 @@ class TexiWikiMarkup (WikiMarkup): | |||
188 | def str_ind(self, elt): | 188 | def str_ind(self, elt): |
189 | return ("@w{ }" * elt['level']) + self.format(elt['content']) | 189 | return ("@w{ }" * elt['level']) + self.format(elt['content']) + '\n' |
190 | 190 | ||
diff --git a/wikimarkup.py b/wikimarkup.py index 9a79d1e..0c6d2f2 100644 --- a/wikimarkup.py +++ b/wikimarkup.py | |||
@@ -186,10 +186,19 @@ class BaseWikiMarkup(object): | |||
186 | if m and line[m.start(0)] != '<': | 186 | if m and line[m.start(0)] != '<': |
187 | if m.group(0)[0] in envtypes and pos < len(line) and line[pos] == ":": | 187 | content = m.group(0) |
188 | yield({ 'type': 'DELIM', | 188 | if content[0] in envtypes: |
189 | 'content': m.group(0), | 189 | t = { 'type': 'DELIM', |
190 | 'continuation': True }) | 190 | 'content': content, |
191 | pos += 1 | 191 | 'continuation': pos < len(line) and line[pos] == ":" } |
192 | if t['continuation']: | ||
193 | t['content'] += t['content'][0] | ||
194 | pos += 1 | ||
195 | |||
196 | yield(t) | ||
197 | |||
198 | while pos < len(line) and line[pos] in [' ', '\t']: | ||
199 | pos += 1 | ||
192 | else: | 200 | else: |
193 | yield({ 'type': 'DELIM', | 201 | yield({ 'type': 'DELIM', |
194 | 'content': m.group(0) }) | 202 | 'content': content, |
203 | 'continuation': False}) | ||
195 | else: | 204 | else: |
@@ -307,4 +316,4 @@ class BaseWikiMarkup(object): | |||
307 | 316 | ||
308 | def peektkn(self): | 317 | def peektkn(self, off=0): |
309 | return self.toklist[self.tokind] | 318 | return self.toklist[self.tokind-off] |
310 | 319 | ||
@@ -606,3 +615,3 @@ class BaseWikiMarkup(object): | |||
606 | elt = self.parse_line() | 615 | elt = self.parse_line() |
607 | if 'continuation' not in tok: | 616 | if not tok['continuation']: |
608 | list.append({ 'type': 'ELT', | 617 | list.append({ 'type': 'ELT', |
@@ -683,6 +692,8 @@ class BaseWikiMarkup(object): | |||
683 | lev = len(tok['content']) | 692 | lev = len(tok['content']) |
693 | if tok['content'][0] == ':': | ||
694 | t = self.peektkn(2) | ||
695 | if not (t['type'] == 'DELIM' and t['content'] == ';'): | ||
696 | return self.parse_indent(lev) | ||
684 | self.ungetkn() | 697 | self.ungetkn() |
685 | return self.parse_env(type, lev) | 698 | return self.parse_env(type, lev) |
686 | elif tok['content'][0] == ":": | ||
687 | return self.parse_indent(len(tok['content'])) | ||
688 | else: | 699 | else: |