Skip to content
Snippets Groups Projects
Commit 2a0314c3 authored by Ximin Luo's avatar Ximin Luo
Browse files

Refactor html-dir presenter to be a class instance, avoiding global state 

No functionality should have been modified by this commit
parent ad28ed35
No related branches found
No related tags found
No related merge requests found
Show whitespace changes
Inline Side-by-side
diffoscope/presenters/html/html.py
+ 406
− 425
View file @ 2a0314c3
......@@ -71,49 +71,6 @@ logger = logging.getLogger(__name__)
re_anchor_prefix = re.compile(r'^[^A-Za-z]')
re_anchor_suffix = re.compile(r'[^A-Za-z-_:\.]')
buf, add_cpt, del_cpt = [], 0, 0
line1, line2, has_internal_linenos = 0, 0, True
hunk_off1, hunk_size1, hunk_off2, hunk_size2 = 0, 0, 0, 0
spl_rows, spl_current_page = 0, 0
spl_print_func, spl_print_ctrl = None, None
class HTMLPresenter(Presenter):
supports_visual_diffs = True
@classmethod
def run(cls, data, difference, parsed_args):
with make_printer(parsed_args.html_output) as fn:
output_html(
difference,
css_url=parsed_args.css_url,
print_func=fn,
)
class HTMLDirectoryPresenter(HTMLPresenter):
@classmethod
def run(cls, data, difference, parsed_args):
output_html_directory(
parsed_args.html_output_directory,
difference,
css_url=parsed_args.css_url,
jquery_url=parsed_args.jquery_url,
)
def new_unified_diff():
global buf, add_cpt, del_cpt
global line1, line2, has_internal_linenos
global hunk_off1, hunk_size1, hunk_off2, hunk_size2
global spl_rows, spl_current_page
global spl_print_func, spl_print_ctrl
buf, add_cpt, del_cpt = [], 0, 0
line1, line2, has_internal_linenos = 0, 0, True
hunk_off1, hunk_size1, hunk_off2, hunk_size2 = 0, 0, 0, 0
spl_rows, spl_current_page = 0, 0
spl_print_func, spl_print_ctrl = None, None
def convert(s, ponct=0, tag=''):
i = 0
......@@ -152,16 +109,97 @@ def convert(s, ponct=0, tag=''):
return t.getvalue()
def output_visual(print_func, visual, parents):
logger.debug('including image for %s', visual.source)
sources = parents + [visual.source]
print_func(u'<div class="difference">')
print_func(u'<div class="diffheader">')
print_func(u'<div class="diffcontrol">⊟</div>')
print_func(u'<div><span class="source">%s</span>'
% html.escape(visual.source))
anchor = escape_anchor('/'.join(sources[1:]))
print_func(
u' <a class="anchor" href="#%s" name="%s">\xb6</a>' % (anchor, anchor))
print_func(u"</div>")
print_func(u"</div>")
print_func(u'<div class="difference">'
u'<img src=\"data:%s,%s\" alt=\"compared images\" /></div>' %
(visual.data_type, visual.content))
print_func(u"</div>", force=True)
def escape_anchor(val):
"""
ID and NAME tokens must begin with a letter ([A-Za-z]) and may be followed
by any number of letters, digits ([0-9]), hyphens ("-"), underscores ("_"),
colons (":"), and periods (".").
"""
for pattern, repl in (
(re_anchor_prefix, 'D'),
(re_anchor_suffix, '-'),
):
val = pattern.sub(repl, val)
return val
def output_header(css_url, print_func):
if css_url:
css_link = '<link href="%s" type="text/css" rel="stylesheet" />' % css_url
else:
css_link = ''
print_func(templates.HEADER % {'title': html.escape(' '.join(sys.argv)),
'favicon': FAVICON_BASE64,
'css_link': css_link,
})
def output_footer(print_func):
print_func(templates.FOOTER % {'version': VERSION}, force=True)
def output_hunk():
spl_print_func(u'<tr class="diffhunk"><td colspan="2">Offset %d, %d lines modified</td>'%(hunk_off1, hunk_size1))
spl_print_func(u'<td colspan="2">Offset %d, %d lines modified</td></tr>\n'%(hunk_off2, hunk_size2))
row_was_output()
@contextlib.contextmanager
def file_printer(directory, filename):
with codecs.open(os.path.join(directory,filename), 'w', encoding='utf-8') as f:
yield f.write
@contextlib.contextmanager
def spl_file_printer(directory, filename):
with codecs.open(os.path.join(directory,filename), 'w', encoding='utf-8') as f:
print_func = f.write
def recording_print_func(s, force=False):
print_func(s)
recording_print_func.bytes_written += len(s)
recording_print_func.bytes_written = 0
yield recording_print_func
def output_line(s1, s2):
global line1, line2, has_internal_linenos
class HTMLPresenter(Presenter):
supports_visual_diffs = True
def __init__(self):
self.new_unified_diff()
def new_unified_diff(self):
self.buf = []
self.add_cpt = 0
self.del_cpt = 0
self.line1 = 0
self.line2 = 0
self.has_internal_linenos = True
self.hunk_off1 = 0
self.hunk_size1 = 0
self.hunk_off2 = 0
self.hunk_size2 = 0
self.spl_rows = 0
self.spl_current_page = 0
self.spl_print_func = None
self.spl_print_ctrl = None
def output_hunk(self):
self.spl_print_func(u'<tr class="diffhunk"><td colspan="2">Offset %d, %d lines modified</td>' % (self.hunk_off1, self.hunk_size1))
self.spl_print_func(u'<td colspan="2">Offset %d, %d lines modified</td></tr>\n' % (self.hunk_off2, self.hunk_size2))
self.row_was_output()
def output_line(self, s1, s2):
orig1 = s1
orig2 = s2
......@@ -184,57 +222,52 @@ def output_line(s1, s2):
type_name = "changed"
s1, s2 = linediff(s1, s2, DIFFON, DIFFOFF)
spl_print_func(u'<tr class="diff%s">' % type_name)
self.spl_print_func(u'<tr class="diff%s">' % type_name)
try:
if s1:
if has_internal_linenos:
spl_print_func(u'<td colspan="2" class="diffpresent">')
if self.has_internal_linenos:
self.spl_print_func(u'<td colspan="2" class="diffpresent">')
else:
spl_print_func(u'<td class="diffline">%d </td>' % line1)
spl_print_func(u'<td class="diffpresent">')
spl_print_func(convert(s1, ponct=1, tag='del'))
spl_print_func(u'</td>')
self.spl_print_func(u'<td class="diffline">%d </td>' % self.line1)
self.spl_print_func(u'<td class="diffpresent">')
self.spl_print_func(convert(s1, ponct=1, tag='del'))
self.spl_print_func(u'</td>')
else:
spl_print_func(u'<td colspan="2">\xa0</td>')
self.spl_print_func(u'<td colspan="2">\xa0</td>')
if s2:
if has_internal_linenos:
spl_print_func(u'<td colspan="2" class="diffpresent">')
if self.has_internal_linenos:
self.spl_print_func(u'<td colspan="2" class="diffpresent">')
else:
spl_print_func(u'<td class="diffline">%d </td>' % line2)
spl_print_func(u'<td class="diffpresent">')
spl_print_func(convert(s2, ponct=1, tag='ins'))
spl_print_func(u'</td>')
self.spl_print_func(u'<td class="diffline">%d </td>' % self.line2)
self.spl_print_func(u'<td class="diffpresent">')
self.spl_print_func(convert(s2, ponct=1, tag='ins'))
self.spl_print_func(u'</td>')
else:
spl_print_func(u'<td colspan="2">\xa0</td>')
self.spl_print_func(u'<td colspan="2">\xa0</td>')
finally:
spl_print_func(u"</tr>\n", force=True)
row_was_output()
self.spl_print_func(u"</tr>\n", force=True)
self.row_was_output()
m = orig1 and re.match(r"^\[ (\d+) lines removed \]$", orig1)
if m:
line1 += int(m.group(1))
self.line1 += int(m.group(1))
elif orig1:
line1 += 1
self.line1 += 1
m = orig2 and re.match(r"^\[ (\d+) lines removed \]$", orig2)
if m:
line2 += int(m.group(1))
self.line2 += int(m.group(1))
elif orig2:
line2 += 1
self.line2 += 1
def empty_buffer():
global buf
global add_cpt
global del_cpt
def empty_buffer(self):
if self.del_cpt == 0 or self.add_cpt == 0:
for l in self.buf:
self.output_line(l[0], l[1])
if del_cpt == 0 or add_cpt == 0:
for l in buf:
output_line(l[0], l[1])
elif del_cpt != 0 and add_cpt != 0:
elif self.del_cpt != 0 and self.add_cpt != 0:
l0, l1 = [], []
for l in buf:
for l in self.buf:
if l[0] != None:
l0.append(l[0])
if l[1] != None:
......@@ -246,241 +279,188 @@ def empty_buffer():
s0 = l0[i]
if i < len(l1):
s1 = l1[i]
output_line(s0, s1)
add_cpt, del_cpt = 0, 0
buf = []
self.output_line(s0, s1)
self.add_cpt = 0
self.del_cpt = 0
self.buf = []
def spl_print_enter(print_context, rotation_params):
def spl_print_enter(self, print_context, rotation_params):
# Takes ownership of print_context
global spl_print_func, spl_print_ctrl
spl_print_ctrl = print_context.__exit__, rotation_params
spl_print_func = print_context.__enter__()
self.spl_print_ctrl = print_context.__exit__, rotation_params
self.spl_print_func = print_context.__enter__()
_, _, css_url = rotation_params
# Print file and table headers
output_header(css_url, spl_print_func)
def spl_had_entered_child():
global spl_print_ctrl, spl_current_page
return spl_print_ctrl and spl_print_ctrl[1] and spl_current_page > 0
def spl_print_exit(*exc_info):
global spl_print_func, spl_print_ctrl
if not spl_had_entered_child(): return False
output_footer(spl_print_func)
_exit, _ = spl_print_ctrl
spl_print_func, spl_print_ctrl = None, None
return _exit(*exc_info)
output_header(css_url, self.spl_print_func)
@contextlib.contextmanager
def spl_file_printer(directory, filename):
with codecs.open(os.path.join(directory,filename), 'w', encoding='utf-8') as f:
print_func = f.write
def recording_print_func(s, force=False):
print_func(s)
recording_print_func.bytes_written += len(s)
recording_print_func.bytes_written = 0
yield recording_print_func
def spl_had_entered_child(self):
return self.spl_print_ctrl and self.spl_print_ctrl[1] and self.spl_current_page > 0
def spl_print_exit(self, *exc_info):
if not self.spl_had_entered_child(): return False
output_footer(self.spl_print_func)
_exit, _ = self.spl_print_ctrl
self.spl_print_func = None
self.spl_print_ctrl = None
return _exit(*exc_info)
def row_was_output():
global spl_print_func, spl_print_ctrl, spl_rows, spl_current_page
spl_rows += 1
_, rotation_params = spl_print_ctrl
def row_was_output(self):
self.spl_rows += 1
_, rotation_params = self.spl_print_ctrl
max_lines = Config().max_diff_block_lines
max_lines_parent = Config().max_diff_block_lines_parent
max_lines_ratio = Config().max_diff_block_lines_html_dir_ratio
max_report_child_size = Config().max_report_child_size
if not rotation_params:
# html-dir single output, don't need to rotate
if spl_rows >= max_lines:
if self.spl_rows >= max_lines:
raise DiffBlockLimitReached()
return
else:
# html-dir output, perhaps need to rotate
directory, mainname, css_url = rotation_params
if spl_rows >= max_lines_ratio * max_lines:
if self.spl_rows >= max_lines_ratio * max_lines:
raise DiffBlockLimitReached()
if spl_current_page == 0: # on parent page
if spl_rows < max_lines_parent:
if self.spl_current_page == 0: # on parent page
if self.spl_rows < max_lines_parent:
return
else: # on child page
# TODO: make this stay below the max, instead of going 1 row over the max
# will require some backtracking...
if spl_print_func.bytes_written < max_report_child_size:
if self.spl_print_func.bytes_written < max_report_child_size:
return
spl_current_page += 1
filename = "%s-%s.html" % (mainname, spl_current_page)
self.spl_current_page += 1
filename = "%s-%s.html" % (mainname, self.spl_current_page)
if spl_current_page > 1:
if self.spl_current_page > 1:
# previous page was a child, close it
spl_print_func(templates.UD_TABLE_FOOTER % {"filename": html.escape(filename), "text": "load diff"}, force=True)
spl_print_exit(None, None, None)
self.spl_print_func(templates.UD_TABLE_FOOTER % {"filename": html.escape(filename), "text": "load diff"}, force=True)
self.spl_print_exit(None, None, None)
# rotate to the next child page
context = spl_file_printer(directory, filename)
spl_print_enter(context, rotation_params)
spl_print_func(templates.UD_TABLE_HEADER)
self.spl_print_enter(context, rotation_params)
self.spl_print_func(templates.UD_TABLE_HEADER)
def output_unified_diff_table(unified_diff, _has_internal_linenos):
global add_cpt, del_cpt
global line1, line2, has_internal_linenos
global hunk_off1, hunk_size1, hunk_off2, hunk_size2
has_internal_linenos = _has_internal_linenos
spl_print_func(templates.UD_TABLE_HEADER)
def output_unified_diff_table(self, unified_diff, has_internal_linenos):
self.has_internal_linenos = has_internal_linenos
self.spl_print_func(templates.UD_TABLE_HEADER)
try:
bytes_processed = 0
for l in unified_diff.splitlines():
bytes_processed += len(l) + 1
m = re.match(r'^--- ([^\s]*)', l)
if m:
empty_buffer()
self.empty_buffer()
continue
m = re.match(r'^\+\+\+ ([^\s]*)', l)
if m:
empty_buffer()
self.empty_buffer()
continue
m = re.match(r"@@ -(\d+),?(\d*) \+(\d+),?(\d*)", l)
if m:
empty_buffer()
self.empty_buffer()
hunk_data = map(lambda x:x=="" and 1 or int(x), m.groups())
hunk_off1, hunk_size1, hunk_off2, hunk_size2 = hunk_data
line1, line2 = hunk_off1, hunk_off2
output_hunk()
self.hunk_off1, self.hunk_size1, self.hunk_off2, self.hunk_size2 = hunk_data
self.line1, self.line2 = self.hunk_off1, self.hunk_off2
self.output_hunk()
continue
if re.match(r'^\[', l):
empty_buffer()
spl_print_func(u'<td colspan="2">%s</td>\n' % l)
self.empty_buffer()
self.spl_print_func(u'<td colspan="2">%s</td>\n' % l)
if re.match(r"^\\ No newline", l):
if hunk_size2 == 0:
buf[-1] = (buf[-1][0], buf[-1][1] + '\n' + l[2:])
if self.hunk_size2 == 0:
self.buf[-1] = (self.buf[-1][0], self.buf[-1][1] + '\n' + l[2:])
else:
buf[-1] = (buf[-1][0] + '\n' + l[2:], buf[-1][1])
self.buf[-1] = (buf[-1][0] + '\n' + l[2:], self.buf[-1][1])
continue
if hunk_size1 <= 0 and hunk_size2 <= 0:
empty_buffer()
if self.hunk_size1 <= 0 and self.hunk_size2 <= 0:
self.empty_buffer()
continue
m = re.match(r"^\+\[ (\d+) lines removed \]$", l)
if m:
add_cpt += int(m.group(1))
hunk_size2 -= int(m.group(1))
buf.append((None, l[1:]))
self.add_cpt += int(m.group(1))
self.hunk_size2 -= int(m.group(1))
self.buf.append((None, l[1:]))
continue
if re.match(r"^\+", l):
add_cpt += 1
hunk_size2 -= 1
buf.append((None, l[1:]))
self.add_cpt += 1
self.hunk_size2 -= 1
self.buf.append((None, l[1:]))
continue
m = re.match(r"^-\[ (\d+) lines removed \]$", l)
if m:
del_cpt += int(m.group(1))
hunk_size1 -= int(m.group(1))
buf.append((l[1:], None))
self.del_cpt += int(m.group(1))
self.hunk_size1 -= int(m.group(1))
self.buf.append((l[1:], None))
continue
if re.match(r"^-", l):
del_cpt += 1
hunk_size1 -= 1
buf.append((l[1:], None))
self.del_cpt += 1
self.hunk_size1 -= 1
self.buf.append((l[1:], None))
continue
if re.match(r"^ ", l) and hunk_size1 and hunk_size2:
empty_buffer()
hunk_size1 -= 1
hunk_size2 -= 1
buf.append((l[1:], l[1:]))
if re.match(r"^ ", l) and self.hunk_size1 and self.hunk_size2:
self.empty_buffer()
self.hunk_size1 -= 1
self.hunk_size2 -= 1
self.buf.append((l[1:], l[1:]))
continue
empty_buffer()
self.empty_buffer()
empty_buffer()
self.empty_buffer()
return True
except DiffBlockLimitReached:
total = len(unified_diff)
bytes_left = total - bytes_processed
frac = bytes_left / total
spl_print_func(
self.spl_print_func(
u'<tr class="error">'
u'<td colspan="4">Max diff block lines reached; %s/%s bytes (%.2f%%) of diff not shown.'
u"</td></tr>" % (bytes_left, total, frac*100), force=True)
return False
except PrintLimitReached:
assert not spl_had_entered_child() # limit reached on the parent page
spl_print_func(u'<tr class="error"><td colspan="4">Max output size reached.</td></tr>', force=True)
assert not self.spl_had_entered_child() # limit reached on the parent page
self.spl_print_func(u'<tr class="error"><td colspan="4">Max output size reached.</td></tr>', force=True)
raise
finally:
spl_print_func(u"</table>", force=True)
self.spl_print_func(u"</table>", force=True)
def output_unified_diff(print_func, css_url, directory, unified_diff, has_internal_linenos):
global spl_print_func, spl_print_ctrl, spl_current_page
new_unified_diff()
def output_unified_diff(self, print_func, css_url, directory, unified_diff, has_internal_linenos):
self.new_unified_diff()
rotation_params = None
if directory:
mainname = hashlib.md5(unified_diff.encode('utf-8')).hexdigest()
rotation_params = directory, mainname, css_url
try:
spl_print_func = print_func
spl_print_ctrl = None, rotation_params
truncated = not output_unified_diff_table(unified_diff, has_internal_linenos)
self.spl_print_func = print_func
self.spl_print_ctrl = None, rotation_params
truncated = not self.output_unified_diff_table(unified_diff, has_internal_linenos)
except:
if not spl_print_exit(*sys.exc_info()): raise
if not self.spl_print_exit(*sys.exc_info()): raise
else:
spl_print_exit(None, None, None)
self.spl_print_exit(None, None, None)
finally:
spl_print_ctrl = None
spl_print_func = None
self.spl_print_ctrl = None
self.spl_print_func = None
if spl_current_page > 0:
noun = "pieces" if spl_current_page > 1 else "piece"
text = "load diff (%s %s%s)" % (spl_current_page, noun, (", truncated" if truncated else ""))
if self.spl_current_page > 0:
noun = "pieces" if self.spl_current_page > 1 else "piece"
text = "load diff (%s %s%s)" % (self.spl_current_page, noun, (", truncated" if truncated else ""))
print_func(templates.UD_TABLE_FOOTER % {"filename": html.escape("%s-1.html" % mainname), "text": text}, force=True)
def output_visual(print_func, visual, parents):
logger.debug('including image for %s', visual.source)
sources = parents + [visual.source]
print_func(u'<div class="difference">')
print_func(u'<div class="diffheader">')
print_func(u'<div class="diffcontrol">⊟</div>')
print_func(u'<div><span class="source">%s</span>'
% html.escape(visual.source))
anchor = escape_anchor('/'.join(sources[1:]))
print_func(
u' <a class="anchor" href="#%s" name="%s">\xb6</a>' % (anchor, anchor))
print_func(u"</div>")
print_func(u"</div>")
print_func(u'<div class="difference">'
u'<img src=\"data:%s,%s\" alt=\"compared images\" /></div>' %
(visual.data_type, visual.content))
print_func(u"</div>", force=True)
def escape_anchor(val):
"""
ID and NAME tokens must begin with a letter ([A-Za-z]) and may be followed
by any number of letters, digits ([0-9]), hyphens ("-"), underscores ("_"),
colons (":"), and periods (".").
"""
for pattern, repl in (
(re_anchor_prefix, 'D'),
(re_anchor_suffix, '-'),
):
val = pattern.sub(repl, val)
return val
def output_difference(difference, print_func, css_url, directory, parents):
def output_difference(self, difference, print_func, css_url, directory, parents):
logger.debug('html output for %s', difference.source1)
sources = parents + [difference.source1]
print_func(u'<div class="difference">')
......@@ -508,31 +488,16 @@ def output_difference(difference, print_func, css_url, directory, parents):
for visual in difference.visuals:
output_visual(print_func, visual, sources)
elif difference.unified_diff:
output_unified_diff(print_func, css_url, directory, difference.unified_diff, difference.has_internal_linenos)
self.output_unified_diff(print_func, css_url, directory, difference.unified_diff, difference.has_internal_linenos)
for detail in difference.details:
output_difference(detail, print_func, css_url, directory, sources)
self.output_difference(detail, print_func, css_url, directory, sources)
except PrintLimitReached:
logger.debug('print limit reached')
raise
finally:
print_func(u"</div>", force=True)
def output_header(css_url, print_func):
if css_url:
css_link = '<link href="%s" type="text/css" rel="stylesheet" />' % css_url
else:
css_link = ''
print_func(templates.HEADER % {'title': html.escape(' '.join(sys.argv)),
'favicon': FAVICON_BASE64,
'css_link': css_link,
})
def output_footer(print_func):
print_func(templates.FOOTER % {'version': VERSION}, force=True)
def output_html(difference, css_url=None, print_func=None):
def output_html(self, difference, css_url=None, print_func=None):
"""
Default presenter, all in one HTML file
"""
......@@ -541,19 +506,26 @@ def output_html(difference, css_url=None, print_func=None):
print_func = create_limited_print_func(print_func, Config().max_report_size)
try:
output_header(css_url, print_func)
output_difference(difference, print_func, css_url, None, [])
self.output_difference(difference, print_func, css_url, None, [])
except PrintLimitReached:
logger.debug('print limit reached')
print_func(u'<div class="error">Max output size reached.</div>',
force=True)
output_footer(print_func)
@contextlib.contextmanager
def file_printer(directory, filename):
with codecs.open(os.path.join(directory,filename), 'w', encoding='utf-8') as f:
yield f.write
@classmethod
def run(cls, data, difference, parsed_args):
with make_printer(parsed_args.html_output) as fn:
cls().output_html(
difference,
css_url=parsed_args.css_url,
print_func=fn,
)
def output_html_directory(directory, difference, css_url=None, jquery_url=None):
class HTMLDirectoryPresenter(HTMLPresenter):
def output_html_directory(self, directory, difference, css_url=None, jquery_url=None):
"""
Multi-file presenter. Writes to a directory, and puts large diff tables
into files of their own.
......@@ -590,7 +562,7 @@ def output_html_directory(directory, difference, css_url=None, jquery_url=None):
print_func = create_limited_print_func(print_func, Config().max_report_size)
try:
output_header(css_url, print_func)
output_difference(difference, print_func, css_url, directory, [])
self.output_difference(difference, print_func, css_url, directory, [])
except PrintLimitReached:
logger.debug('print limit reached')
print_func(u'<div class="error">Max output size reached.</div>',
......@@ -598,3 +570,12 @@ def output_html_directory(directory, difference, css_url=None, jquery_url=None):
if jquery_url:
print_func(templates.SCRIPTS % {'jquery_url': html.escape(jquery_url)}, force=True)
output_footer(print_func)
@classmethod
def run(cls, data, difference, parsed_args):
cls().output_html_directory(
parsed_args.html_output_directory,
difference,
css_url=parsed_args.css_url,
jquery_url=parsed_args.jquery_url,
)
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment