DOCX: Handle hyperlinks created as fields

See https://bugs.launchpad.net/calibre/+bug/1196728 for an example.

src/calibre/ebooks/docx/fields.py

@@ -0,0 +1,106 @@
+#!/usr/bin/env python
+# vim:fileencoding=utf-8
+from __future__ import (unicode_literals, division, absolute_import,
+                        print_function)
+
+__license__ = 'GPL v3'
+__copyright__ = '2013, Kovid Goyal <kovid at kovidgoyal.net>'
+
+import re
+
+from calibre.ebooks.docx.names import XPath, get
+
+class Field(object):
+
+    def __init__(self, start):
+        self.start = start
+        self.end = None
+        self.contents = []
+        self.instructions = []
+
+    def add_instr(self, elem):
+        raw = elem.text
+        if not raw:
+            return
+        name, rest = raw.strip().partition(' ')[0::2]
+        self.instructions.append((name, rest.strip()))
+
+WORD, FLAG = 0, 1
+scanner = re.Scanner([
+    (r'\\\S{1}', lambda s, t: (t, FLAG)),  # A flag of the form \x
+    (r'"[^"]*"', lambda s, t: (t[1:-1], WORD)),  # Quoted word
+    (r'[^\s\\"]\S*', lambda s, t: (t, WORD)),  # A non-quoted word, must not start with a backslash or a space or a quote
+    (r'\s+', None),
+], flags=re.DOTALL)
+
+
+def parse_hyperlink(raw, log):
+    ans = {}
+    last_option = None
+    for token, token_type in scanner.scan(raw)[0]:
+        if not ans:
+            if token_type is not WORD:
+                log('Invalid hyperlink, first token is not a URL (%s)' % raw)
+                return ans
+            ans['url'] = token
+        if token_type is FLAG:
+            last_option = {'l':'anchor', 'm':'image-map', 'n':'target', 'o':'title', 't':'target'}.get(token[1], None)
+            if last_option is not None:
+                ans[last_option] = None
+        elif token_type is WORD:
+            if last_option is not None:
+                ans[last_option] = token
+    return ans
+
+
+class Fields(object):
+
+    def __init__(self):
+        self.fields = []
+
+    def __call__(self, doc, log):
+        stack = []
+        for elem in XPath(
+            '//*[name()="w:p" or name()="w:r" or name()="w:instrText" or (name()="w:fldChar" and (@w:fldCharType="begin" or @w:fldCharType="end"))]')(doc):
+            if elem.tag.endswith('}fldChar'):
+                typ = get(elem, 'w:fldCharType')
+                if typ == 'begin':
+                    stack.append(Field(elem))
+                    self.fields.append(stack[-1])
+                else:
+                    try:
+                        stack.pop().end = elem
+                    except IndexError:
+                        pass
+            elif elem.tag.endswith('}instrText'):
+                if stack:
+                    stack[-1].add_instr(elem)
+            else:
+                if stack:
+                    stack[-1].contents.append(elem)
+
+        # Parse hyperlink fields
+        self.hyperlink_fields = []
+        for field in self.fields:
+            if len(field.instructions) == 1 and field.instructions[0][0] == 'HYPERLINK':
+                hl = parse_hyperlink(field.instructions[0][1], log)
+                if hl:
+                    if 'target' in hl and hl['target'] is None:
+                        hl['target'] = '_blank'
+                    all_runs = []
+                    current_runs = []
+                    # We only handle spans in a single paragraph
+                    # being wrapped in <a>
+                    for x in field.contents:
+                        if x.tag.endswith('}p'):
+                            if current_runs:
+                                all_runs.append(current_runs)
+                            current_runs = []
+                        elif x.tag.endswith('}r'):
+                            current_runs.append(x)
+                    if current_runs:
+                        all_runs.append(current_runs)
+                    for runs in all_runs:
+                        self.hyperlink_fields.append((hl, runs))
+
+

src/calibre/ebooks/docx/to_html.py

@@ -26,6 +26,7 @@ from calibre.ebooks.docx.footnotes import Footnotes
 from calibre.ebooks.docx.cleanup import cleanup_markup
 from calibre.ebooks.docx.theme import Theme
 from calibre.ebooks.docx.toc import create_toc
+from calibre.ebooks.docx.fields import Fields
 from calibre.ebooks.metadata.opf2 import OPFCreator
 from calibre.utils.localization import canonicalize_lang, lang_as_iso639_1
 
@@ -52,6 +53,7 @@ class Convert(object):
         self.body = BODY()
         self.theme = Theme()
         self.tables = Tables()
+        self.fields = Fields()
         self.styles = Styles(self.tables)
         self.images = Images()
         self.object_map = OrderedDict()
@@ -79,6 +81,7 @@ class Convert(object):
     def __call__(self):
         doc = self.docx.document
         relationships_by_id, relationships_by_type = self.docx.document_relationships
+        self.fields(doc, self.log)
         self.read_styles(relationships_by_type)
         self.images(relationships_by_id)
         self.layers = OrderedDict()
@@ -396,6 +399,25 @@ class Convert(object):
             # hrefs that point nowhere give epubcheck a hernia. The element
             # should be styled explicitly by Word anyway.
             # span.set('href', '#')
+        rmap = {v:k for k, v in self.object_map.iteritems()}
+        for hyperlink, runs in self.fields.hyperlink_fields:
+            spans = [rmap[r] for r in runs if r in rmap]
+            if not spans:
+                continue
+            if len(spans) > 1:
+                span = self.wrap_elems(spans, SPAN())
+            span.tag = 'a'
+            tgt = hyperlink.get('target', None)
+            if tgt:
+                span.set('target', tgt)
+            tt = hyperlink.get('title', None)
+            if tt:
+                span.set('title', tt)
+            url = hyperlink['url']
+            if url in self.anchor_map:
+                span.set('href', '#' + self.anchor_map[url])
+                continue
+            span.set('href', url)
 
     def convert_run(self, run):
         ans = SPAN()