TristanThrush
commited on
Commit
•
5f2fd21
1
Parent(s):
ec1654a
modified to take advantage of multiple CPUs locally
Browse files- README.md +8 -7
- wikipedia.py +58 -45
README.md
CHANGED
@@ -626,6 +626,10 @@ configs:
|
|
626 |
|
627 |
# Dataset Card for Wikipedia
|
628 |
|
|
|
|
|
|
|
|
|
629 |
## Table of Contents
|
630 |
- [Dataset Description](#dataset-description)
|
631 |
- [Dataset Summary](#dataset-summary)
|
@@ -667,10 +671,10 @@ markdown and unwanted sections (references, etc.).
|
|
667 |
|
668 |
The articles are parsed using the ``mwparserfromhell`` tool.
|
669 |
|
670 |
-
To load this dataset you need to install
|
671 |
|
672 |
```
|
673 |
-
pip install
|
674 |
```
|
675 |
|
676 |
Then, you can load any subset of Wikipedia per language and per date this way:
|
@@ -678,11 +682,8 @@ Then, you can load any subset of Wikipedia per language and per date this way:
|
|
678 |
```python
|
679 |
from datasets import load_dataset
|
680 |
|
681 |
-
load_dataset("wikipedia", language="sw", date="20220120"
|
682 |
```
|
683 |
-
where you can pass as `beam_runner` any Apache Beam supported runner for (distributed) data processing
|
684 |
-
(see [here](https://beam.apache.org/documentation/runners/capability-matrix/)).
|
685 |
-
Pass "DirectRunner" to run it on your machine.
|
686 |
|
687 |
You can find the full list of languages and dates [here](https://dumps.wikimedia.org/backup-index.html).
|
688 |
|
@@ -856,4 +857,4 @@ the text.
|
|
856 |
|
857 |
### Contributions
|
858 |
|
859 |
-
Thanks to [@lewtun](https://github.com/lewtun), [@mariamabarham](https://github.com/mariamabarham), [@thomwolf](https://github.com/thomwolf), [@lhoestq](https://github.com/lhoestq), [@patrickvonplaten](https://github.com/patrickvonplaten) for adding this dataset.
|
|
|
626 |
|
627 |
# Dataset Card for Wikipedia
|
628 |
|
629 |
+
This repo is a fork of the original Hugging Face Wikipedia repo [here](https://huggingface.co/datasets/wikipedia).
|
630 |
+
The difference is that this fork does away with the need for `apache-beam`, and this fork is also very fast if you have a lot of CPUs on your machine.
|
631 |
+
It will also use all CPUs available to create a clean Wikipedia pretraining dataset. It takes less than an hour to process all of English wikipedia on a GCP n1-standard-96.
|
632 |
+
|
633 |
## Table of Contents
|
634 |
- [Dataset Description](#dataset-description)
|
635 |
- [Dataset Summary](#dataset-summary)
|
|
|
671 |
|
672 |
The articles are parsed using the ``mwparserfromhell`` tool.
|
673 |
|
674 |
+
To load this dataset you need to install ``mwparserfromhell`` first:
|
675 |
|
676 |
```
|
677 |
+
pip install mwparserfromhell
|
678 |
```
|
679 |
|
680 |
Then, you can load any subset of Wikipedia per language and per date this way:
|
|
|
682 |
```python
|
683 |
from datasets import load_dataset
|
684 |
|
685 |
+
load_dataset("wikipedia", language="sw", date="20220120")
|
686 |
```
|
|
|
|
|
|
|
687 |
|
688 |
You can find the full list of languages and dates [here](https://dumps.wikimedia.org/backup-index.html).
|
689 |
|
|
|
857 |
|
858 |
### Contributions
|
859 |
|
860 |
+
Thanks to [@lewtun](https://github.com/lewtun), [@mariamabarham](https://github.com/mariamabarham), [@thomwolf](https://github.com/thomwolf), [@lhoestq](https://github.com/lhoestq), [@patrickvonplaten](https://github.com/patrickvonplaten) for adding this dataset.
|
wikipedia.py
CHANGED
@@ -23,6 +23,9 @@ import json
|
|
23 |
import re
|
24 |
import xml.etree.cElementTree as etree
|
25 |
from urllib.parse import quote
|
|
|
|
|
|
|
26 |
|
27 |
import datasets
|
28 |
|
@@ -904,7 +907,7 @@ class WikipediaConfig(datasets.BuilderConfig):
|
|
904 |
_DATE = "20220301"
|
905 |
|
906 |
|
907 |
-
class Wikipedia(datasets.
|
908 |
"""Wikipedia dataset."""
|
909 |
|
910 |
# Use mirror (your.org) to avoid download caps.
|
@@ -934,7 +937,7 @@ class Wikipedia(datasets.BeamBasedBuilder):
|
|
934 |
citation=_CITATION,
|
935 |
)
|
936 |
|
937 |
-
def _split_generators(self, dl_manager
|
938 |
def _base_url(lang):
|
939 |
return _BASE_URL_TMPL.format(lang=lang.replace("-", "_"), date=self.config.date)
|
940 |
|
@@ -963,9 +966,10 @@ class Wikipedia(datasets.BeamBasedBuilder):
|
|
963 |
xml_urls.append(_base_url(lang) + fname)
|
964 |
|
965 |
# Use dictionary since testing mock always returns the same result.
|
|
|
|
|
966 |
downloaded_files = dl_manager.download({"xml": xml_urls})
|
967 |
-
|
968 |
-
downloaded_files = dl_manager.ship_files_with_pipeline(downloaded_files, pipeline)
|
969 |
|
970 |
return [
|
971 |
datasets.SplitGenerator( # pylint:disable=g-complex-comprehension
|
@@ -973,43 +977,40 @@ class Wikipedia(datasets.BeamBasedBuilder):
|
|
973 |
)
|
974 |
]
|
975 |
|
976 |
-
def
|
977 |
-
|
978 |
-
import apache_beam as beam
|
979 |
-
import mwparserfromhell
|
980 |
|
981 |
def _extract_content(filepath):
|
982 |
"""Extracts article content from a single WikiMedia XML file."""
|
983 |
logger.info("generating examples from = %s", filepath)
|
984 |
-
|
985 |
-
|
986 |
-
|
987 |
-
|
988 |
-
|
989 |
-
|
990 |
-
|
991 |
-
|
992 |
-
|
993 |
-
|
994 |
-
|
995 |
-
|
996 |
-
|
997 |
-
|
998 |
-
|
999 |
-
|
1000 |
-
elem.clear()
|
1001 |
-
continue
|
1002 |
-
|
1003 |
-
raw_content = elem.find(f"./{namespace}revision/{namespace}text").text
|
1004 |
elem.clear()
|
|
|
|
|
|
|
|
|
1005 |
|
1006 |
-
|
1007 |
-
|
1008 |
-
|
1009 |
-
continue
|
1010 |
|
1011 |
-
|
1012 |
-
|
1013 |
|
1014 |
def _clean_content(inputs, language):
|
1015 |
"""Cleans raw wikicode to extract text."""
|
@@ -1017,28 +1018,40 @@ class Wikipedia(datasets.BeamBasedBuilder):
|
|
1017 |
try:
|
1018 |
text = _parse_and_clean_wikicode(raw_content, parser=mwparserfromhell, language=language)
|
1019 |
except (mwparserfromhell.parser.ParserError) as e:
|
1020 |
-
beam.metrics.Metrics.counter(language, "parser-error").inc()
|
1021 |
logger.error("mwparserfromhell ParseError: %s", e)
|
1022 |
return
|
1023 |
|
1024 |
if not text:
|
1025 |
-
beam.metrics.Metrics.counter(language, "empty-clean-examples").inc()
|
1026 |
return
|
1027 |
|
1028 |
url = _construct_url(title, language)
|
1029 |
|
1030 |
-
|
1031 |
|
1032 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1033 |
|
1034 |
-
|
1035 |
-
|
1036 |
-
|
1037 |
-
|
1038 |
-
|
1039 |
-
|
1040 |
-
|
|
|
|
|
1041 |
|
|
|
1042 |
|
1043 |
def _parse_and_clean_wikicode(raw_content, parser, language):
|
1044 |
"""Strips formatting and unwanted sections from raw page content."""
|
|
|
23 |
import re
|
24 |
import xml.etree.cElementTree as etree
|
25 |
from urllib.parse import quote
|
26 |
+
import mwparserfromhell
|
27 |
+
from multiprocessing import Process, Manager
|
28 |
+
from tqdm import tqdm
|
29 |
|
30 |
import datasets
|
31 |
|
|
|
907 |
_DATE = "20220301"
|
908 |
|
909 |
|
910 |
+
class Wikipedia(datasets.GeneratorBasedBuilder):
|
911 |
"""Wikipedia dataset."""
|
912 |
|
913 |
# Use mirror (your.org) to avoid download caps.
|
|
|
937 |
citation=_CITATION,
|
938 |
)
|
939 |
|
940 |
+
def _split_generators(self, dl_manager):
|
941 |
def _base_url(lang):
|
942 |
return _BASE_URL_TMPL.format(lang=lang.replace("-", "_"), date=self.config.date)
|
943 |
|
|
|
966 |
xml_urls.append(_base_url(lang) + fname)
|
967 |
|
968 |
# Use dictionary since testing mock always returns the same result.
|
969 |
+
|
970 |
+
print("Dowloading Wikipedia dump")
|
971 |
downloaded_files = dl_manager.download({"xml": xml_urls})
|
972 |
+
print("Finished downloading Wikipedia dump")
|
|
|
973 |
|
974 |
return [
|
975 |
datasets.SplitGenerator( # pylint:disable=g-complex-comprehension
|
|
|
977 |
)
|
978 |
]
|
979 |
|
980 |
+
def _generate_examples(self, filepaths, language, no_labels=False):
|
981 |
+
|
|
|
|
|
982 |
|
983 |
def _extract_content(filepath):
|
984 |
"""Extracts article content from a single WikiMedia XML file."""
|
985 |
logger.info("generating examples from = %s", filepath)
|
986 |
+
content = []
|
987 |
+
f = bz2.BZ2File(filename=filepath)
|
988 |
+
# Workaround due to: https://github.com/tensorflow/tensorflow/issues/33563
|
989 |
+
utf_f = codecs.getreader("utf-8")(f)
|
990 |
+
context = etree.iterparse(utf_f, events=("end",))
|
991 |
+
for unused_event, elem in context:
|
992 |
+
if not elem.tag.endswith("page"):
|
993 |
+
continue
|
994 |
+
namespace = elem.tag[:-4]
|
995 |
+
title = elem.find(f"./{namespace}title").text
|
996 |
+
ns = elem.find(f"./{namespace}ns").text
|
997 |
+
id_ = elem.find(f"./{namespace}id").text
|
998 |
+
red_ = elem.find(f"./{namespace}redirect")
|
999 |
+
|
1000 |
+
# Filter pages that are not in the "main" namespace.
|
1001 |
+
if ns != "0":
|
|
|
|
|
|
|
|
|
1002 |
elem.clear()
|
1003 |
+
continue
|
1004 |
+
|
1005 |
+
raw_content = elem.find(f"./{namespace}revision/{namespace}text").text
|
1006 |
+
elem.clear()
|
1007 |
|
1008 |
+
# Filter redirects.
|
1009 |
+
if raw_content is None or red_ is not None:
|
1010 |
+
continue
|
|
|
1011 |
|
1012 |
+
content.append((id_, title, raw_content))
|
1013 |
+
return content
|
1014 |
|
1015 |
def _clean_content(inputs, language):
|
1016 |
"""Cleans raw wikicode to extract text."""
|
|
|
1018 |
try:
|
1019 |
text = _parse_and_clean_wikicode(raw_content, parser=mwparserfromhell, language=language)
|
1020 |
except (mwparserfromhell.parser.ParserError) as e:
|
|
|
1021 |
logger.error("mwparserfromhell ParseError: %s", e)
|
1022 |
return
|
1023 |
|
1024 |
if not text:
|
|
|
1025 |
return
|
1026 |
|
1027 |
url = _construct_url(title, language)
|
1028 |
|
1029 |
+
return id_, {"id": id_, "url": url, "title": title, "text": text}
|
1030 |
|
1031 |
+
print("Parsing and cleaning Wikipedia examples")
|
1032 |
+
with Manager() as manager:
|
1033 |
+
examples = manager.list()
|
1034 |
+
processes = []
|
1035 |
+
for filepath in filepaths:
|
1036 |
+
def parse_and_clean(examples):
|
1037 |
+
content = _extract_content(filepath)
|
1038 |
+
for obj in tqdm(content):
|
1039 |
+
examples.append(_clean_content(obj, language=language))
|
1040 |
+
p = Process(target=parse_and_clean, args=(examples,))
|
1041 |
+
p.start()
|
1042 |
+
processes.append(p)
|
1043 |
|
1044 |
+
for p in processes:
|
1045 |
+
p.join()
|
1046 |
+
|
1047 |
+
print("Parsed and cleaned Wikipedia examples")
|
1048 |
+
|
1049 |
+
for example in examples:
|
1050 |
+
if example is not None:
|
1051 |
+
yield example
|
1052 |
+
|
1053 |
|
1054 |
+
|
1055 |
|
1056 |
def _parse_and_clean_wikicode(raw_content, parser, language):
|
1057 |
"""Strips formatting and unwanted sections from raw page content."""
|