Skip to content

Commit 6ee1c5a

Browse files
author
Mark Pilgrim
committed
Refactored error messages to codes defined in constants.py
--HG-- extra : convert_revision : svn%3Aacbfec75-9323-0410-a652-858a13e371e0/trunk%40962
1 parent eea8b56 commit 6ee1c5a

File tree

6 files changed

+425
-204
lines changed

6 files changed

+425
-204
lines changed

‎parse.py‎

Lines changed: 3 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -14,6 +14,7 @@
1414
#END RELEASE
1515
from html5lib import html5parser, liberalxmlparser
1616
from html5lib import treebuilders, serializer, treewalkers
17+
from html5lib import constants
1718

1819
def parse():
1920
optParser = getOptParser()
@@ -100,8 +101,8 @@ def printOutput(parser, document, opts):
100101
if not text.endswith('\n'): sys.stdout.write('\n')
101102
if opts.error:
102103
errList=[]
103-
for pos, message in parser.errors:
104-
errList.append("Line %i Col %i"%pos + " " + message)
104+
for pos, errorcode, datavars in parser.errors:
105+
errList.append("Line %i Col %i"%pos + " " + constants.E.get(errorcode, 'Unknown error "%s"' % errorcode) % datavars)
105106
sys.stdout.write("\nParse errors:\n" + "\n".join(errList)+"\n")
106107

107108
def getOptParser():

‎src/html5lib/constants.py‎

Lines changed: 227 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,5 @@
1-
import string
1+
import string, gettext
2+
_ = gettext.gettext
23

34
try:
45
frozenset
@@ -9,6 +10,230 @@
910

1011
EOF = None
1112

13+
E = {
14+
"null-character":
15+
_(u"Null character in input stream, replaced with U+FFFD."),
16+
"incorrectly-placed-solidus":
17+
_(u"Solidus (/) incorrectly placed in tag."),
18+
"incorrect-cr-newline-entity":
19+
_(u"Incorrect CR newline entity, replaced with LF."),
20+
"illegal-windows-1252-entity":
21+
_(u"Entity used with illegal number (windows-1252 reference)."),
22+
"cant-convert-numeric-entity":
23+
_(u"Numeric entity couldn't be converted to character "
24+
u"(codepoint U+%(charAsInt)08x)."),
25+
"illegal-codepoint-for-numeric-entity":
26+
_(u"Numeric entity represents an illegal codepoint: "
27+
u"U+%(charAsInt)08x."),
28+
"numeric-entity-without-semicolon":
29+
_(u"Numeric entity didn't end with ';'."),
30+
"expected-numeric-entity-but-got-eof":
31+
_(u"Numeric entity expected. Got end of file instead."),
32+
"expected-numeric-entity":
33+
_(u"Numeric entity expected but none found."),
34+
"named-entity-without-semicolon":
35+
_(u"Named entity didn't end with ';'."),
36+
"expected-named-entity":
37+
_(u"Named entity expected. Got none."),
38+
"attributes-in-end-tag":
39+
_(u"End tag contains unexpected attributes."),
40+
"expected-tag-name-but-got-right-bracket":
41+
_(u"Expected tag name. Got '>' instead."),
42+
"expected-tag-name-but-got-question-mark":
43+
_(u"Expected tag name. Got '?' instead. (HTML doesn't "
44+
u"support processing instructions.)"),
45+
"expected-tag-name":
46+
_(u"Expected tag name. Got something else instead"),
47+
"expected-closing-tag-but-got-right-bracket":
48+
_(u"Expected closing tag. Got '>' instead. Ignoring '</>'."),
49+
"expected-closing-tag-but-got-eof":
50+
_(u"Expected closing tag. Unexpected end of file."),
51+
"expected-closing-tag-but-got-char":
52+
_(u"Expected closing tag. Unexpected character '%(data)s' found."),
53+
"eof-in-tag-name":
54+
_(u"Unexpected end of file in the tag name."),
55+
"expected-attribute-name-but-got-eof":
56+
_(u"Unexpected end of file. Expected attribute name instead."),
57+
"eof-in-attribute-name":
58+
_(u"Unexpected end of file in attribute name."),
59+
"duplicate-attribute":
60+
_(u"Dropped duplicate attribute on tag."),
61+
"expected-end-of-tag-name-but-got-eof":
62+
_(u"Unexpected end of file. Expected = or end of tag."),
63+
"expected-attribute-value-but-got-eof":
64+
_(u"Unexpected end of file. Expected attribute value."),
65+
"eof-in-attribute-value-double-quote":
66+
_(u"Unexpected end of file in attribute value (\")."),
67+
"eof-in-attribute-value-single-quote":
68+
_(u"Unexpected end of file in attribute value (')."),
69+
"eof-in-attribute-value-no-quotes":
70+
_(u"Unexpected end of file in attribute value."),
71+
"expected-dashes-or-doctype":
72+
_(u"Expected '--' or 'DOCTYPE'. Not found."),
73+
"incorrect-comment":
74+
_(u"Incorrect comment."),
75+
"eof-in-comment":
76+
_(u"Unexpected end of file in comment."),
77+
"eof-in-comment-end-dash":
78+
_(u"Unexpected end of file in comment (-)"),
79+
"unexpected-dash-after-double-dash-in-comment":
80+
_(u"Unexpected '-' after '--' found in comment."),
81+
"eof-in-comment-double-dash":
82+
_(u"Unexpected end of file in comment (--)."),
83+
"unexpected-char-in-comment":
84+
_(u"Unexpected character in comment found."),
85+
"need-space-after-doctype":
86+
_(u"No space after literal string 'DOCTYPE'."),
87+
"expected-doctype-name-but-got-right-bracket":
88+
_(u"Unexpected > character. Expected DOCTYPE name."),
89+
"expected-doctype-name-but-got-eof":
90+
_(u"Unexpected end of file. Expected DOCTYPE name."),
91+
"eof-in-doctype-name":
92+
_(u"Unexpected end of file in DOCTYPE name."),
93+
"eof-in-doctype":
94+
_(u"Unexpected end of file in DOCTYPE."),
95+
"expected-space-or-right-bracket-in-doctype":
96+
_(u"Expected space or '>'. Got '%(data)s'"),
97+
"unexpected-end-of-doctype":
98+
_(u"Unexpected end of DOCTYPE."),
99+
"unexpected-char-in-doctype":
100+
_(u"Unexpected character in DOCTYPE."),
101+
"eof-in-bogus-doctype":
102+
_(u"Unexpected end of file in bogus doctype."),
103+
"eof-in-innerhtml":
104+
_(u"XXX innerHTML EOF"),
105+
"unexpected-doctype":
106+
_(u"Unexpected DOCTYPE. Ignored."),
107+
"non-html-root":
108+
_(u"html needs to be the first start tag."),
109+
"expected-doctype-but-got-eof":
110+
_(u"Unexpected End of file. Expected DOCTYPE."),
111+
"unknown-doctype":
112+
_(u"Erroneous DOCTYPE."),
113+
"expected-doctype-but-got-chars":
114+
_(u"Unexpected non-space characters. Expected DOCTYPE."),
115+
"expected-doctype-but-got-start-tag":
116+
_(u"Unexpected start tag (%(name)s). Expected DOCTYPE."),
117+
"expected-doctype-but-got-end-tag":
118+
_(u"Unexpected end tag (%(name)s). Expected DOCTYPE."),
119+
"end-tag-after-implied-root":
120+
_(u"Unexpected end tag (%(name)s) after the (implied) root element."),
121+
"expected-named-closing-tag-but-got-eof":
122+
_(u"Unexpected end of file. Expected end tag (%(name)s)."),
123+
"two-heads-are-not-better-than-one":
124+
_(u"Unexpected start tag head in existing head. Ignored."),
125+
"unexpected-end-tag":
126+
_(u"Unexpected end tag (%(name)s). Ignored."),
127+
"unexpected-start-tag-out-of-my-head":
128+
_(u"Unexpected start tag (%(name)s) that can be in head. Moved."),
129+
"unexpected-start-tag":
130+
_(u"Unexpected start tag (%(name)s)."),
131+
"missing-end-tag":
132+
_(u"Missing end tag (%(name)s)."),
133+
"missing-end-tags":
134+
_(u"Missing end tags (%(name)s)."),
135+
"unexpected-start-tag-implies-end-tag":
136+
_(u"Unexpected start tag (%(startName)s) "
137+
u"implies end tag (%(endName)s)."),
138+
"unexpected-start-tag-treated-as":
139+
_(u"Unexpected start tag (%(originalName)s). Treated as %(newName)s."),
140+
"deprecated-tag":
141+
_(u"Unexpected start tag %(name)s. Don't use it!"),
142+
"unexpected-start-tag-ignored":
143+
_(u"Unexpected start tag %(name)s. Ignored."),
144+
"expected-one-end-tag-but-got-another":
145+
_(u"Unexpected end tag (%(gotName)s). "
146+
u"Missing end tag (%(expectedName)s)."),
147+
"end-tag-too-early":
148+
_(u"End tag (%(name)s) seen too early. Expected other end tag."),
149+
"end-tag-too-early-named":
150+
_(u"Unexpected end tag (%(gotName)s). Expected end tag (%(expectedName)s)."),
151+
"end-tag-too-early-ignored":
152+
_(u"End tag (%(name)s) seen too early. Ignored."),
153+
"adoption-agency-1.1":
154+
_(u"End tag (%(name)s) violates step 1, "
155+
u"paragraph 1 of the adoption agency algorithm."),
156+
"adoption-agency-1.2":
157+
_(u"End tag (%(name)s) violates step 1, "
158+
u"paragraph 2 of the adoption agency algorithm."),
159+
"adoption-agency-1.3":
160+
_(u"End tag (%(name)s) violates step 1, "
161+
u"paragraph 3 of the adoption agency algorithm."),
162+
"unexpected-end-tag-treated-as":
163+
_(u"Unexpected end tag (%(originalName)s). Treated as %(newName)s."),
164+
"no-end-tag":
165+
_(u"This element (%(name)s) has no end tag."),
166+
"unexpected-implied-end-tag-in-table":
167+
_(u"Unexpected implied end tag (%(name)s) in the table phase."),
168+
"unexpected-implied-end-tag-in-table-body":
169+
_(u"Unexpected implied end tag (%(name)s) in the table body phase."),
170+
"unexpected-char-implies-table-voodoo":
171+
_(u"Unexpected non-space characters in "
172+
u"table context caused voodoo mode."),
173+
"unexpected-start-tag-implies-table-voodoo":
174+
_(u"Unexpected start tag (%(name)s) in "
175+
u"table context caused voodoo mode."),
176+
"unexpected-end-tag-implies-table-voodoo":
177+
_(u"Unexpected end tag (%(name)s) in "
178+
u"table context caused voodoo mode."),
179+
"unexpected-cell-in-table-body":
180+
_(u"Unexpected table cell start tag (%(name)s) "
181+
u"in the table body phase."),
182+
"unexpected-cell-end-tag":
183+
_(u"Got table cell end tag (%(name)s) "
184+
u"while required end tags are missing."),
185+
"unexpected-end-tag-in-table-body":
186+
_(u"Unexpected end tag (%(name)s) in the table body phase. Ignored."),
187+
"unexpected-implied-end-tag-in-table-row":
188+
_(u"Unexpected implied end tag (%(name)s) in the table row phase."),
189+
"unexpected-end-tag-in-table-row":
190+
_(u"Unexpected end tag (%(name)s) in the table row phase. Ignored."),
191+
"unexpected-select-in-select":
192+
_(u"Unexpected select start tag in the select phase "
193+
u"implies select start tag."),
194+
"unexpected-start-tag-in-select":
195+
_(u"Unexpected start tag token (%(name)s in the select phase. "
196+
u"Ignored."),
197+
"unexpected-end-tag-in-select":
198+
_(u"Unexpected end tag (%(name)s) in the select phase. Ignored."),
199+
"unexpected-char-after-body":
200+
_(u"Unexpected non-space characters in the after body phase."),
201+
"unexpected-start-tag-after-body":
202+
_(u"Unexpected start tag token (%(name)s)"
203+
u" in the after body phase."),
204+
"unexpected-end-tag-after-body":
205+
_(u"Unexpected end tag token (%(name)s)"
206+
u" in the after body phase."),
207+
"unexpected-char-in-frameset":
208+
_(u"Unepxected characters in the frameset phase. Characters ignored."),
209+
"unexpected-start-tag-in-frameset":
210+
_(u"Unexpected start tag token (%(name)s)"
211+
u" in the frameset phase. Ignored."),
212+
"unexpected-frameset-in-frameset-innerhtml":
213+
_(u"Unexpected end tag token (frameset) "
214+
u"in the frameset phase (innerHTML)."),
215+
"unexpected-end-tag-in-frameset":
216+
_(u"Unexpected end tag token (%(name)s)"
217+
u" in the frameset phase. Ignored."),
218+
"unexpected-char-after-frameset":
219+
_(u"Unexpected non-space characters in the "
220+
u"after frameset phase. Ignored."),
221+
"unexpected-start-tag-after-frameset":
222+
_(u"Unexpected start tag (%(name)s)"
223+
u" in the after frameset phase. Ignored."),
224+
"unexpected-end-tag-after-frameset":
225+
_(u"Unexpected end tag (%(name)s)"
226+
u" in the after frameset phase. Ignored."),
227+
"expected-eof-but-got-char":
228+
_(u"Unexpected non-space characters. Expected end of file."),
229+
"expected-eof-but-got-start-tag":
230+
_(u"Unexpected start tag (%(name)s)"
231+
u". Expected end of file."),
232+
"expected-eof-but-got-end-tag":
233+
_(u"Unexpected end tag (%(name)s)"
234+
u". Expected end of file."),
235+
}
236+
12237
contentModelFlags = {
13238
"PCDATA":0,
14239
"RCDATA":1,
@@ -813,4 +1038,4 @@
8131038
"windows-1258",
8141039
"tis-620",
8151040
"hz-gb-2312",
816-
))
1041+
))

0 commit comments

Comments
 (0)