Skip to content

pick_parallel_docs

Pick out parallel files to prestable/converted inside a corpus directory.

ParallelPicker

Pick valid parallel files from converted xml files.

Attributes:

Name Type Description
vcs versioncontrol.vcs

version control client for the corpus directory
language1_dir str

the directory where converted files of language1 are found
parallel_language str

three character long language code
minratio float

the lowest diff in percent between wordcount in the language1 and parallel document that is accepted
maxratio float

the highest diff in percent between wordcount in the language1 and parallel document that is accepted
poor_ratio list of tuple

each tuple contains a pair of filename paths and the word count ratio of this pair.
counter defaultdict(int

count various things.
Source code in /home/anders/projects/CorpusTools/corpustools/pick_parallel_docs.py 
 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
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
class ParallelPicker:
    """Pick valid parallel files from converted xml files.

    Attributes:
        vcs (versioncontrol.vcs): version control client for the corpus
            directory
        language1_dir (str): the directory where converted files of language1
            are found
        parallel_language (str): three character long language code
        minratio (float): the lowest diff in percent between wordcount in the
            language1 and parallel document that is accepted
        maxratio (float): the highest diff in percent between wordcount in the
            language1 and parallel document that is accepted
        poor_ratio (list of tuple): each tuple contains a pair of filename
            paths and the word count ratio of this pair.
        counter (defaultdict(int)): count various things.
    """

    def __init__(self, language1_dir, parallel_language, minratio, maxratio):
        """Initialise the ParallelPicker class.

        Args:
            language1_dir (str): the directory where the lang1 files exist.
            parallel_language (str): the parallel language where the lang2
                files exist.
            minratio (int): the minimum acceptable ratio of sentences
                between two parallel documents
            maxratio (int): the maximum acceptable ratio of sentences
                between two parallel documents
        """
        self.vcs = versioncontrol.vcs(language1_dir[: language1_dir.find("converted/")])
        self.language1_dir = language1_dir
        self.calculate_language1(language1_dir)
        self.parallel_language = parallel_language
        self.minratio = float(minratio)
        self.maxratio = float(maxratio)
        self.poor_ratio = []
        self.counter = defaultdict(int)

    def calculate_language1(self, language1_dir):
        """The language is the part after 'converted/'."""
        converted_pos = language1_dir.find("converted/")
        part_after_converted = language1_dir[converted_pos + len("converted/") :]

        if part_after_converted.find("/") == -1:
            self.language1 = part_after_converted
        else:
            self.language1 = part_after_converted[: part_after_converted.find("/")]

    def find_lang1_files(self):
        """Find the language1 files.

        Yields:
            (corpusxmlfile.CorpusXMLFile): Language files
        """
        for root, _, files in os.walk(self.language1_dir):
            for lang1_file in files:
                if lang1_file.endswith(".xml"):
                    yield corpusxmlfile.CorpusXMLFile(os.path.join(root, lang1_file))

    def has_parallel(self, language1_file):
        """Check if the given file has a parallel file.

        Args:
            language1_file (corpusxmlfile.CorpusXMLFile): The first file of a
                parallel pair.

        Returns:
            (bool): True if there is a parallel file, False otherwise
        """
        parallel_name = language1_file.get_parallel_filename(self.parallel_language)
        return parallel_name is not None and os.path.isfile(parallel_name)

    @staticmethod
    def has_sufficient_words(file1, file2):
        """Check if the given file contains more words than the threshold.

        Args:
            file1 (corpusxmlfile.CorpusXMLFile): The first file of a parallel
                pair.
            file2 (corpusxmlfile.CorpusXMLFile): The second file of a parallel
                pair.

        Returns:
            (bool): True if true, False if not
        """
        threshold = 30
        return (
            file1.word_count is not None
            and int(file1.word_count) > threshold
            and file2.word_count is not None
            and int(file2.word_count) > threshold
        )

    def has_sufficient_ratio(self, file1, file2):
        """See if the ratio of words is good enough.

        Args:
            file1 (corpusxmlfile.CorpusXMLFile): The first file of a parallel
                pair.
            file2 (corpusxmlfile.CorpusXMLFile): The second file of a parallel
                pair.

        Returns:
            (True|None): True if good enough, None if not
        """
        ratio = float(file1.word_count) / float(file2.word_count) * 100
        if self.minratio < ratio < self.maxratio:
            return True
        else:
            self.poor_ratio.append(
                (file1.name, file1.word_count, file2.name, file2.word_count, ratio)
            )

    def copy_file(self, xml_file):
        """Copy xml_file to prestable/converted.

        Args:
            xml_file (corpusxmlfile.CorpusXMLFile): the file that should be
                copied to prestable/converted.
        """
        prestable_dir = xml_file.dirname.replace("converted", "prestable/converted")

        if not os.path.isdir(prestable_dir):
            with util.ignored(OSError):
                os.makedirs(prestable_dir)

        prestable_name = os.path.join(prestable_dir, xml_file.basename)
        shutil.copy(xml_file.name, prestable_dir)
        self.vcs.add(prestable_name)

        return prestable_name

    def is_valid_pair(self, file1, file2):
        """Check if file1 and file2 is a valid parallel pair.

        Args:
            file1 (corpusxmlfile.CorpusXMLFile): The first file of a parallel
                pair.
            file2 (corpusxmlfile.CorpusXMLFile): The second file of a parallel
                pair.

        Returns:
            (bool): True if valid, False if not
        """
        return self.has_sufficient_words(file1, file2) and self.has_sufficient_ratio(
            file1, file2
        )

    def valid_parallels(self):
        """Pick valid parallel file pairs.

        Yields:
            (tuple[corpusxmlfile.CorpusXMLFile]): valid file pairs
        """
        for language1_file in self.find_lang1_files():
            if self.has_parallel(language1_file):
                self.counter["has_parallel"] += 1
                parallel_file = corpusxmlfile.CorpusXMLFile(
                    language1_file.get_parallel_filename(self.parallel_language)
                )
                if self.is_valid_pair(language1_file, parallel_file):
                    self.counter["good_parallel"] += 1
                    yield (language1_file, parallel_file)

    def copy_valid_parallels(self):
        """Copy valid parallel files from converted to prestable/converted."""
        for file1, file2 in self.valid_parallels():
            prestable_name = self.copy_file(file1)
            self.copy_file(file2)
            outfile = parallelize.parallelise_file(
                prestable_name,
                file2.lang,
                dictionary=None,
                quiet=True,
                aligner="tca2",
                stdout=False,
                force=True,
            )

            self.vcs.add(outfile)
            self.vcs.add(outfile + ".html")

    def print_report(self):
        for poor_ratio in self.poor_ratio:
            print("{}: {}\n{}: {}\nratio: {}\n".format(*poor_ratio))

        for key, value in self.counter.items():
            print(key, value)

__init__(language1_dir, parallel_language, minratio, maxratio)

Initialise the ParallelPicker class.

Parameters:

Name Type Description Default
language1_dir str

the directory where the lang1 files exist.

 required
parallel_language str

the parallel language where the lang2 files exist.

 required
minratio int

the minimum acceptable ratio of sentences between two parallel documents

 required
maxratio int

the maximum acceptable ratio of sentences between two parallel documents

 required
Source code in /home/anders/projects/CorpusTools/corpustools/pick_parallel_docs.py 
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
def __init__(self, language1_dir, parallel_language, minratio, maxratio):
    """Initialise the ParallelPicker class.

    Args:
        language1_dir (str): the directory where the lang1 files exist.
        parallel_language (str): the parallel language where the lang2
            files exist.
        minratio (int): the minimum acceptable ratio of sentences
            between two parallel documents
        maxratio (int): the maximum acceptable ratio of sentences
            between two parallel documents
    """
    self.vcs = versioncontrol.vcs(language1_dir[: language1_dir.find("converted/")])
    self.language1_dir = language1_dir
    self.calculate_language1(language1_dir)
    self.parallel_language = parallel_language
    self.minratio = float(minratio)
    self.maxratio = float(maxratio)
    self.poor_ratio = []
    self.counter = defaultdict(int)

calculate_language1(language1_dir)

The language is the part after 'converted/'.

Source code in /home/anders/projects/CorpusTools/corpustools/pick_parallel_docs.py 
75
76
77
78
79
80
81
82
83
def calculate_language1(self, language1_dir):
    """The language is the part after 'converted/'."""
    converted_pos = language1_dir.find("converted/")
    part_after_converted = language1_dir[converted_pos + len("converted/") :]

    if part_after_converted.find("/") == -1:
        self.language1 = part_after_converted
    else:
        self.language1 = part_after_converted[: part_after_converted.find("/")]

copy_file(xml_file)

Copy xml_file to prestable/converted.

Parameters:

Name Type Description Default
xml_file corpusxmlfile.CorpusXMLFile

the file that should be copied to prestable/converted.

 required
Source code in /home/anders/projects/CorpusTools/corpustools/pick_parallel_docs.py 
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
def copy_file(self, xml_file):
    """Copy xml_file to prestable/converted.

    Args:
        xml_file (corpusxmlfile.CorpusXMLFile): the file that should be
            copied to prestable/converted.
    """
    prestable_dir = xml_file.dirname.replace("converted", "prestable/converted")

    if not os.path.isdir(prestable_dir):
        with util.ignored(OSError):
            os.makedirs(prestable_dir)

    prestable_name = os.path.join(prestable_dir, xml_file.basename)
    shutil.copy(xml_file.name, prestable_dir)
    self.vcs.add(prestable_name)

    return prestable_name

copy_valid_parallels()

Copy valid parallel files from converted to prestable/converted.

Source code in /home/anders/projects/CorpusTools/corpustools/pick_parallel_docs.py 
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
def copy_valid_parallels(self):
    """Copy valid parallel files from converted to prestable/converted."""
    for file1, file2 in self.valid_parallels():
        prestable_name = self.copy_file(file1)
        self.copy_file(file2)
        outfile = parallelize.parallelise_file(
            prestable_name,
            file2.lang,
            dictionary=None,
            quiet=True,
            aligner="tca2",
            stdout=False,
            force=True,
        )

        self.vcs.add(outfile)
        self.vcs.add(outfile + ".html")

find_lang1_files()

Find the language1 files.

Yields:

Type Description
corpusxmlfile.CorpusXMLFile

Language files
Source code in /home/anders/projects/CorpusTools/corpustools/pick_parallel_docs.py 
85
86
87
88
89
90
91
92
93
94
def find_lang1_files(self):
    """Find the language1 files.

    Yields:
        (corpusxmlfile.CorpusXMLFile): Language files
    """
    for root, _, files in os.walk(self.language1_dir):
        for lang1_file in files:
            if lang1_file.endswith(".xml"):
                yield corpusxmlfile.CorpusXMLFile(os.path.join(root, lang1_file))

has_parallel(language1_file)

Check if the given file has a parallel file.

Parameters:

Name Type Description Default
language1_file corpusxmlfile.CorpusXMLFile

The first file of a parallel pair.

 required

Returns:

Type Description
bool

True if there is a parallel file, False otherwise
Source code in /home/anders/projects/CorpusTools/corpustools/pick_parallel_docs.py 
 96
 97
 98
 99
100
101
102
103
104
105
106
107
def has_parallel(self, language1_file):
    """Check if the given file has a parallel file.

    Args:
        language1_file (corpusxmlfile.CorpusXMLFile): The first file of a
            parallel pair.

    Returns:
        (bool): True if there is a parallel file, False otherwise
    """
    parallel_name = language1_file.get_parallel_filename(self.parallel_language)
    return parallel_name is not None and os.path.isfile(parallel_name)

has_sufficient_ratio(file1, file2)

See if the ratio of words is good enough.

Parameters:

Name Type Description Default
file1 corpusxmlfile.CorpusXMLFile

The first file of a parallel pair.

 required
file2 corpusxmlfile.CorpusXMLFile

The second file of a parallel pair.

 required

Returns:

Type Description
True | None

True if good enough, None if not
Source code in /home/anders/projects/CorpusTools/corpustools/pick_parallel_docs.py 
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
def has_sufficient_ratio(self, file1, file2):
    """See if the ratio of words is good enough.

    Args:
        file1 (corpusxmlfile.CorpusXMLFile): The first file of a parallel
            pair.
        file2 (corpusxmlfile.CorpusXMLFile): The second file of a parallel
            pair.

    Returns:
        (True|None): True if good enough, None if not
    """
    ratio = float(file1.word_count) / float(file2.word_count) * 100
    if self.minratio < ratio < self.maxratio:
        return True
    else:
        self.poor_ratio.append(
            (file1.name, file1.word_count, file2.name, file2.word_count, ratio)
        )

has_sufficient_words(file1, file2) staticmethod

Check if the given file contains more words than the threshold.

Parameters:

Name Type Description Default
file1 corpusxmlfile.CorpusXMLFile

The first file of a parallel pair.

 required
file2 corpusxmlfile.CorpusXMLFile

The second file of a parallel pair.

 required

Returns:

Type Description
bool

True if true, False if not
Source code in /home/anders/projects/CorpusTools/corpustools/pick_parallel_docs.py 
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
@staticmethod
def has_sufficient_words(file1, file2):
    """Check if the given file contains more words than the threshold.

    Args:
        file1 (corpusxmlfile.CorpusXMLFile): The first file of a parallel
            pair.
        file2 (corpusxmlfile.CorpusXMLFile): The second file of a parallel
            pair.

    Returns:
        (bool): True if true, False if not
    """
    threshold = 30
    return (
        file1.word_count is not None
        and int(file1.word_count) > threshold
        and file2.word_count is not None
        and int(file2.word_count) > threshold
    )

is_valid_pair(file1, file2)

Check if file1 and file2 is a valid parallel pair.

Parameters:

Name Type Description Default
file1 corpusxmlfile.CorpusXMLFile

The first file of a parallel pair.

 required
file2 corpusxmlfile.CorpusXMLFile

The second file of a parallel pair.

 required

Returns:

Type Description
bool

True if valid, False if not
Source code in /home/anders/projects/CorpusTools/corpustools/pick_parallel_docs.py 
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
def is_valid_pair(self, file1, file2):
    """Check if file1 and file2 is a valid parallel pair.

    Args:
        file1 (corpusxmlfile.CorpusXMLFile): The first file of a parallel
            pair.
        file2 (corpusxmlfile.CorpusXMLFile): The second file of a parallel
            pair.

    Returns:
        (bool): True if valid, False if not
    """
    return self.has_sufficient_words(file1, file2) and self.has_sufficient_ratio(
        file1, file2
    )

valid_parallels()

Pick valid parallel file pairs.

Yields:

Type Description
tuple[corpusxmlfile.CorpusXMLFile]

valid file pairs
Source code in /home/anders/projects/CorpusTools/corpustools/pick_parallel_docs.py 
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
def valid_parallels(self):
    """Pick valid parallel file pairs.

    Yields:
        (tuple[corpusxmlfile.CorpusXMLFile]): valid file pairs
    """
    for language1_file in self.find_lang1_files():
        if self.has_parallel(language1_file):
            self.counter["has_parallel"] += 1
            parallel_file = corpusxmlfile.CorpusXMLFile(
                language1_file.get_parallel_filename(self.parallel_language)
            )
            if self.is_valid_pair(language1_file, parallel_file):
                self.counter["good_parallel"] += 1
                yield (language1_file, parallel_file)

main()

Copy valid parallel pairs from converted to prestable/converted.

Source code in /home/anders/projects/CorpusTools/corpustools/pick_parallel_docs.py 
259
260
261
262
263
264
265
266
267
def main():
    """Copy valid parallel pairs from converted to prestable/converted."""
    args = parse_options()

    picker = ParallelPicker(
        args.language1_dir, args.parallel_language, args.minratio, args.maxratio
    )
    picker.copy_valid_parallels()
    picker.print_report()

parse_options()

Parse the commandline options.

Returns:

Type Description
argparse.Namespace

the parsed commandline arguments
Source code in /home/anders/projects/CorpusTools/corpustools/pick_parallel_docs.py 
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
def parse_options():
    """Parse the commandline options.

    Returns:
        (argparse.Namespace): the parsed commandline arguments
    """
    parser = argparse.ArgumentParser(
        parents=[argparse_version.parser],
        description="Pick out parallel files from converted to " "prestable/converted.",
    )

    parser.add_argument(
        "language1_dir", help="directory where the files of language1 exist"
    )
    parser.add_argument(
        "-p",
        "--parallel_language",
        dest="parallel_language",
        help="The language where we would like to find " "parallel documents",
        required=True,
    )
    parser.add_argument(
        "--minratio", dest="minratio", help="The minimum ratio", required=True
    )
    parser.add_argument(
        "--maxratio", dest="maxratio", help="The maximum ratio", required=True
    )

    args = parser.parse_args()
    return args