def test_process_html_unzipped(self):
proc = MetaProcessor(options={'oocp-out-fmt': 'html',
'meta-procord': 'unzip,oocp'})
self.resultpath, metadata = proc.process(self.input)
assert os.path.isfile(self.resultpath)
assert metadata['error'] is False and metadata['oocp_status'] == 0
assert open(self.resultpath, 'r').read().startswith('<!DOCTYPE')
assert self.resultpath.endswith('sample.html')
def test_process_with_errors(self):
proc = MetaProcessor(options={'meta-procord': 'error'})
self.resultpath, metadata = proc.process(self.input)
assert self.resultpath is None
assert metadata == {
'error': True,
'error-descr': 'Intentional error. Please ignore',
}
def test_process_with_errors(self, workdir):
proc = MetaProcessor(options={'meta-procord': 'error'})
resultpath, metadata = proc.process(str(workdir / "src" /
"sample.txt"))
assert resultpath is None
assert metadata == {
'error': True,
'error-descr': 'Intentional error. Please ignore',
}
def test_process_with_errors(self, workdir):
proc = MetaProcessor(options={'meta-procord': 'error'})
resultpath, metadata = proc.process(
str(workdir / "src" / "sample.txt"))
assert resultpath is None
assert metadata == {
'error': True,
'error-descr': 'Intentional error. Please ignore',
}
def test_process_xhtml_unzipped(self, workdir):
proc = MetaProcessor(options={'oocp-out-fmt': 'xhtml',
'meta-procord': 'unzip,oocp'})
resultpath, metadata = proc.process(
str(workdir / "src" / "sample.txt"))
assert os.path.isfile(resultpath)
assert metadata['error'] is False and metadata['oocp_status'] == 0
assert open(resultpath, 'r').read().startswith('<?xml ')
assert resultpath.endswith('sample.html')
def test_options_as_strings(self):
proc = MetaProcessor(options={'meta.procord': 'oocp, oocp'})
result = proc.get_options_as_string()
assert result == (
"css_cleaner_minified=Truehtml_cleaner_fix_heading_numbers=True"
"html_cleaner_fix_image_links=Truehtml_cleaner_fix_sd_fields=True"
"meta_processor_order=('unzip', 'oocp', 'tidy', 'html_cleaner', "
"'css_cleaner', 'zip')oocp_hostname=localhost"
"oocp_output_format=htmloocp_pdf_tagged=False"
"oocp_pdf_version=Falseoocp_port=2002")
def test_process_html_unzipped(self, workdir):
proc = MetaProcessor(options={
'oocp-out-fmt': 'html',
'meta-procord': 'unzip,oocp'
})
resultpath, metadata = proc.process(str(workdir / "src" /
"sample.txt"))
assert os.path.isfile(resultpath)
assert metadata['error'] is False and metadata['oocp_status'] == 0
assert open(resultpath, 'r').read().startswith('<!DOCTYPE')
assert resultpath.endswith('sample.html')
def convert_doc(src_doc, options, cache_dir):
"""Convert `src_doc` according to the other parameters.
`src_doc` is the path to the source document. `options` is a dict
of options for processing, passed to the processors.
`cache_dir` may be ``None`` in which no caching is requested
during processing.
Generates a converted representation of `src_doc` by calling
:class:`ulif.openoffice.processor.MetaProcessor` with `options` as
parameters.
Afterwards the conversion result is stored in cache (if
allowed/possible) for speedup of upcoming requests.
Returns a triple:
``(<PATH>, <CACHE_KEY>, <METADATA>)``
where ``<PATH>`` is the path to the resulting document,
``<CACHE_KEY>`` an identifier (string) to retrieve a generated doc
from cache on future requests, and ``<METADATA>`` is a dict of values
returned during request (and set by the document processors,
notably setting the `error` keyword).
If errors happen or caching is disabled, ``<CACHE_KEY>`` is
``None``.
"""
result_path = None
cache_key = None
repr_key = get_marker(options) # Create unique marker out of options
metadata = dict(error=False)
# Generate result
input_copy_dir = tempfile.mkdtemp()
input_copy = os.path.join(input_copy_dir, os.path.basename(src_doc))
shutil.copy2(src_doc, input_copy)
try:
proc = MetaProcessor(options=options) # Removes original doc
result_path, metadata = proc.process(input_copy)
except Exception as exc:
shutil.rmtree(input_copy_dir)
raise exc
error_state = metadata.get('error', False)
if cache_dir and not error_state and result_path is not None:
# Cache away generated doc
cache_key = CacheManager(cache_dir).register_doc(
src_doc, result_path, repr_key)
return result_path, cache_key, metadata
def convert_doc(src_doc, options, cache_dir):
"""Convert `src_doc` according to the other parameters.
`src_doc` is the path to the source document. `options` is a dict
of options for processing, passed to the processors.
`cache_dir` may be ``None`` in which no caching is requested
during processing.
Generates a converted representation of `src_doc` by calling
:class:`ulif.openoffice.processor.MetaProcessor` with `options` as
parameters.
Afterwards the conversion result is stored in cache (if
allowed/possible) for speedup of upcoming requests.
Returns a triple:
``(<PATH>, <CACHE_KEY>, <METADATA>)``
where ``<PATH>`` is the path to the resulting document,
``<CACHE_KEY>`` an identifier (string) to retrieve a generated doc
from cache on future requests, and ``<METADATA>`` is a dict of values
returned during request (and set by the document processors,
notably setting the `error` keyword).
If errors happen or caching is disabled, ``<CACHE_KEY>`` is
``None``.
"""
result_path = None
cache_key = None
repr_key = get_marker(options) # Create unique marker out of options
metadata = dict(error=False)
# Generate result
input_copy_dir = tempfile.mkdtemp()
input_copy = os.path.join(input_copy_dir, os.path.basename(src_doc))
shutil.copy2(src_doc, input_copy)
try:
proc = MetaProcessor(options=options) # Removes original doc
result_path, metadata = proc.process(input_copy)
except Exception as exc:
shutil.rmtree(input_copy_dir)
raise exc
error_state = metadata.get('error', False)
if cache_dir and not error_state and result_path is not None:
# Cache away generated doc
cache_key = CacheManager(cache_dir).register_doc(
src_doc, result_path, repr_key)
return result_path, cache_key, metadata
def test_options_as_strings(self):
proc = MetaProcessor(options={'meta.procord': 'oocp, oocp'})
result = proc.get_options_as_string()
assert result == (
"css_cleaner_minified=True"
"css_cleaner_prettify_html=False"
"html_cleaner_fix_heading_numbers=True"
"html_cleaner_fix_image_links=True"
"html_cleaner_fix_sd_fields=True"
"meta_processor_order=('unzip', 'oocp', 'tidy', 'html_cleaner', "
"'css_cleaner', 'zip')"
"oocp_hostname=localhost"
"oocp_output_format=html"
"oocp_pdf_tagged=False"
"oocp_pdf_version=False"
"oocp_port=2002")
def test_ignored_options(self):
# We ignore keys not in default dict
proc = MetaProcessor(options={'meta-foo': '12'})
assert 'meta-foo' not in proc.options.keys()
def test_process_default(self, workdir):
proc = MetaProcessor(options={})
resultpath, metadata = proc.process(
str(workdir / "src" / "sample.txt"))
assert metadata['error'] is False and metadata['oocp_status'] == 0
assert resultpath.endswith('sample.html.zip')
def test_build_pipeline_empty_elements(self):
proc = MetaProcessor(options={'meta-procord': 'oocp,,,oocp'})
result = proc._build_pipeline()
assert result == (OOConvProcessor, OOConvProcessor)
def test_build_pipeline_empty(self):
proc = MetaProcessor(options={'meta-procord': ''})
result = proc._build_pipeline()
assert result == ()
def test_build_pipeline_single(self):
proc = MetaProcessor(options={'meta-procord': 'oocp'})
result = proc._build_pipeline()
assert result == (OOConvProcessor,)
def test_options_invalid(self):
# Make sure that invalid options lead to exceptions
with pytest.raises(ArgumentParserError):
MetaProcessor(options={'meta-procord': 'oop,nonsense'})
def test_avail_processors(self):
# Make sure processors defined via entry points are found
proc = MetaProcessor(options={'meta-procord': 'oocp, oocp'})
assert proc.avail_procs['oocp'] is OOConvProcessor
assert len(proc.avail_procs.items()) > 0
def test_process_default(self):
proc = MetaProcessor(options={})
self.resultpath, metadata = proc.process(self.input)
assert metadata['error'] is False and metadata['oocp_status'] == 0
assert self.resultpath.endswith('sample.html.zip')
def test_build_pipeline_single(self):
proc = MetaProcessor(options={'meta-procord': 'oocp'})
result = proc._build_pipeline()
assert result == (OOConvProcessor, )
def test_build_pipeline_empty(self):
proc = MetaProcessor(options={'meta-procord': ''})
result = proc._build_pipeline()
assert result == ()
def test_no_options(self):
# We cope with no options set
proc = MetaProcessor()
assert 'meta_processor_order' in proc.options.keys()
def test_build_pipeline_empty_elements(self):
proc = MetaProcessor(options={'meta-procord': 'oocp,,,oocp'})
result = proc._build_pipeline()
assert result == (OOConvProcessor, OOConvProcessor)
def test_process_default(self, workdir):
proc = MetaProcessor(options={})
resultpath, metadata = proc.process(str(workdir / "src" /
"sample.txt"))
assert metadata['error'] is False and metadata['oocp_status'] == 0
assert resultpath.endswith('sample.html.zip')
def test_non_meta_options(self):
# We ignore options not determined for the meta processor
proc = MetaProcessor(options={'foo.bar': '12'})
assert 'bar' not in proc.options.keys()
def test_option_set(self):
# We respect options set if available in the defaults dict
proc = MetaProcessor(options={'meta-procord': 'oocp,oocp'})
assert proc.options['meta_processor_order'] == ('oocp', 'oocp')
