1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
|
#!/usr/bin/python3
import sys, re
import bz2, html, urllib.parse
import sqlite3
usageInfo = f"usage: {sys.argv[0]}\n"
usageInfo += "For a set of page-ids, looks up their content in an enwiki dump,\n"
usageInfo += "trying to get infobox image filenames, adding info to an sqlite db.\n"
if len(sys.argv) > 1:
print(usageInfo, file=sys.stderr)
sys.exit(1)
def getInputPageIds():
pageIds = set()
dbCon = sqlite3.connect("../data.db")
dbCur = dbCon.cursor()
for (pageId,) in dbCur.execute("SELECT wiki_id from descs"):
pageIds.add(pageId)
dbCon.close()
return pageIds
dumpFile = "enwiki-20220501-pages-articles-multistream.xml.bz2"
indexDb = "dumpIndex.db"
imgDb = "enwikiImgs.db" # Output db
idLineRegex = re.compile(r"<id>(.*)</id>")
imageLineRegex = re.compile(r".*\| *image *= *([^|]*)")
bracketImageRegex = re.compile(r"\[\[(File:[^|]*).*]]")
imageNameRegex = re.compile(r".*\.(jpg|jpeg|png|gif|tiff|tif)", flags=re.IGNORECASE)
cssImgCropRegex = re.compile(r"{{css image crop\|image *= *(.*)", flags=re.IGNORECASE)
# Open dbs
indexDbCon = sqlite3.connect(indexDb)
indexDbCur = indexDbCon.cursor()
imgDbCon = sqlite3.connect(imgDb)
imgDbCur = imgDbCon.cursor()
# Create image-db table
imgDbCur.execute("CREATE TABLE page_imgs (page_id INT PRIMARY KEY, img_name TEXT)")
imgDbCur.execute("CREATE INDEX page_imgs_idx ON page_imgs(img_name)")
# Get input pageIds
print("Getting input page-ids", file=sys.stderr)
pageIds = getInputPageIds()
# Get page-id dump-file offsets
print("Getting dump-file offsets", file=sys.stderr)
offsetToPageids = {}
offsetToEnd = {}
iterNum = 0
for pageId in pageIds:
iterNum += 1
if iterNum % 1e4 == 0:
print(f"At iteration {iterNum}", file=sys.stderr)
#
query = "SELECT offset, next_offset FROM offsets WHERE id = ?"
row = indexDbCur.execute(query, (pageId,)).fetchone()
if row == None:
print(f"WARNING: Page id {pageId} not found", file=sys.stderr)
continue
(chunkOffset, endOffset) = row
offsetToEnd[chunkOffset] = endOffset
if chunkOffset not in offsetToPageids:
offsetToPageids[chunkOffset] = []
offsetToPageids[chunkOffset].append(pageId)
print(f"Found {len(offsetToEnd)} chunks to check", file=sys.stderr)
# Look through dump file, jumping to chunks containing relevant pages
print("Reading through dump file", file=sys.stderr)
def getImageName(content):
""" Given an array of text-content lines, returns an image-filename, or None """
for line in content:
match = imageLineRegex.match(line)
if match != None:
imageName = match.group(1).strip()
if imageName == "":
return None
imageName = html.unescape(imageName)
# Account for {{...
if imageName.startswith("{"):
match = cssImgCropRegex.match(imageName)
if match == None:
return None
imageName = match.group(1)
# Account for [[File:...|...]]
if imageName.startswith("["):
match = bracketImageRegex.match(imageName)
if match == None:
return None
imageName = match.group(1)
# Account for <!--
if imageName.find("<!--") != -1:
return None
# Remove an initial 'File:'
if imageName.startswith("File:"):
imageName = imageName[5:]
# Remove an initial 'Image:'
if imageName.startswith("Image:"):
imageName = imageName[6:]
# Check for extension
match = imageNameRegex.match(imageName)
if match != None:
imageName = match.group(0)
imageName = urllib.parse.unquote(imageName)
imageName = html.unescape(imageName) # Intentionally unescaping again (handles some odd cases)
imageName = imageName.replace("_", " ")
return imageName
# Skip lines like: | image = <imagemap>
return None
# Doesn't try and find images in outside-infobox [[File:...]] and <imagemap> sections
return None
with open(dumpFile, mode='rb') as file:
iterNum = 0
for (pageOffset, endOffset) in offsetToEnd.items():
iterNum += 1
if iterNum % 100 == 0:
print(f"At iteration {iterNum}", file=sys.stderr)
#
pageIds = offsetToPageids[pageOffset]
# Jump to chunk
file.seek(pageOffset)
compressedData = file.read(None if endOffset == -1 else endOffset - pageOffset)
data = bz2.BZ2Decompressor().decompress(compressedData).decode()
# Look in chunk for pages
lines = data.splitlines()
lineIdx = 0
while lineIdx < len(lines):
# Look for <page>
if lines[lineIdx].lstrip() != "<page>":
lineIdx += 1
continue
# Check page id
lineIdx += 3
idLine = lines[lineIdx].lstrip()
match = idLineRegex.fullmatch(idLine)
if match == None or int(match.group(1)) not in pageIds:
lineIdx += 1
continue
pageId = int(match.group(1))
lineIdx += 1
# Look for <text> in <page>
foundText = False
while lineIdx < len(lines):
if not lines[lineIdx].lstrip().startswith("<text "):
lineIdx += 1
continue
foundText = True
# Get text content
content = []
line = lines[lineIdx]
content.append(line[line.find(">") + 1:])
lineIdx += 1
foundTextEnd = False
while lineIdx < len(lines):
line = lines[lineIdx]
if not line.endswith("</text>"):
content.append(line)
lineIdx += 1
continue
foundTextEnd = True
content.append(line[:line.rfind("</text>")])
# Look for image-filename
imageName = getImageName(content)
if imageName != None:
imgDbCur.execute("INSERT into page_imgs VALUES (?, ?)", (pageId, imageName))
break
if not foundTextEnd:
print(f"Did not find </text> for page id {pageId}", file=sys.stderr)
break
if not foundText:
print(f"Did not find <text> for page id {pageId}", file=sys.stderr)
# Close dbs
indexDbCon.close()
imgDbCon.commit()
imgDbCon.close()
|
