Merge branch 'release/4.5'

Author: hbredin

doc/source/changelog.rst

@@ -2,6 +2,13 @@
 Changelog
 #########
 
+Version 4.5 (2022-08-24)
+~~~~~~~~~~~~~~~~~~~~~~~~
+
+- feat: add Annotation.write_lab method (@FrancescoBonzi)
+- feat: add Annotation.to_{rttm|lab} and Timeline.to_uem serializers (@juanmc2005)
+- setup: exclude tests from package (@kimdwkimdw)
+
 Version 4.4 (2022-03-09)
 ~~~~~~~~~~~~~~~~~~~~~~~~
 

pyannote/core/annotation.py

@@ -187,7 +187,7 @@ def __init__(self, uri: Optional[str] = None, modality: Optional[str] = None):
         # key: label
         # value: timeline
         self._labels: Dict[Label, Timeline] = {}
-        self._labelNeedsUpdate: [Label, bool] = {}
+        self._labelNeedsUpdate: Dict[Label, bool] = {}
 
         # timeline meant to store all annotated segments
         self._timeline: Timeline = None
@@ -364,19 +364,14 @@ def __contains__(self, included: Union[Segment, Timeline]):
         """
         return included in self.get_timeline(copy=False)
 
-    def write_rttm(self, file: TextIO):
-        """Dump annotation to file using RTTM format
-
-        Parameters
-        ----------
-        file : file object
+    def _iter_rttm(self) -> Iterator[Text]:
+        """Generate lines for an RTTM file for this annotation
 
-        Usage
-        -----
-        >>> with open('file.rttm', 'w') as file:
-        ...     annotation.write_rttm(file)
+        Returns
+        -------
+        iterator: Iterator[str]
+            An iterator over RTTM text lines
         """
-
         uri = self.uri if self.uri else "<NA>"
         if isinstance(uri, Text) and " " in uri:
             msg = (
@@ -391,10 +386,76 @@ def write_rttm(self, file: TextIO):
                     f'containing spaces (got: "{label}").'
                 )
                 raise ValueError(msg)
-            line = (
+            yield (
                 f"SPEAKER {uri} 1 {segment.start:.3f} {segment.duration:.3f} "
                 f"<NA> <NA> {label} <NA> <NA>\n"
             )
+
+    def to_rttm(self) -> Text:
+        """Serialize annotation as a string using RTTM format
+
+        Returns
+        -------
+        serialized: str
+            RTTM string
+        """
+        return "".join([line for line in self._iter_rttm()])
+
+    def write_rttm(self, file: TextIO):
+        """Dump annotation to file using RTTM format
+
+        Parameters
+        ----------
+        file : file object
+
+        Usage
+        -----
+        >>> with open('file.rttm', 'w') as file:
+        ...     annotation.write_rttm(file)
+        """
+        for line in self._iter_rttm():
+            file.write(line)
+
+    def _iter_lab(self) -> Iterator[Text]:
+        """Generate lines for a LAB file for this annotation
+
+        Returns
+        -------
+        iterator: Iterator[str]
+            An iterator over LAB text lines
+        """
+        for segment, _, label in self.itertracks(yield_label=True):
+            if isinstance(label, Text) and " " in label:
+                msg = (
+                    f"Space-separated LAB file format does not allow labels "
+                    f'containing spaces (got: "{label}").'
+                )
+                raise ValueError(msg)
+            yield f"{segment.start:.3f} {segment.start + segment.duration:.3f} {label}\n"
+
+    def to_lab(self) -> Text:
+        """Serialize annotation as a string using LAB format
+
+        Returns
+        -------
+        serialized: str
+            LAB string
+        """
+        return "".join([line for line in self._iter_lab()])
+
+    def write_lab(self, file: TextIO):
+        """Dump annotation to file using LAB format
+
+        Parameters
+        ----------
+        file : file object
+
+        Usage
+        -----
+        >>> with open('file.lab', 'w') as file:
+        ...     annotation.write_lab(file)
+        """
+        for line in self._iter_lab():
             file.write(line)
 
     def crop(self, support: Support, mode: CropMode = "intersection") -> "Annotation":

pyannote/core/timeline.py

@@ -1069,6 +1069,32 @@ def to_annotation(self,
 
         return annotation
 
+    def _iter_uem(self) -> Iterator[Text]:
+        """Generate lines for a UEM file for this timeline
+
+        Returns
+        -------
+        iterator: Iterator[str]
+            An iterator over UEM text lines
+        """
+        uri = self.uri if self.uri else "<NA>"
+        if isinstance(uri, Text) and ' ' in uri:
+            msg = (f'Space-separated UEM file format does not allow file URIs '
+                   f'containing spaces (got: "{uri}").')
+            raise ValueError(msg)
+        for segment in self:
+            yield f"{uri} 1 {segment.start:.3f} {segment.end:.3f}\n"
+
+    def to_uem(self) -> Text:
+        """Serialize timeline as a string using UEM format
+
+        Returns
+        -------
+        serialized: str
+            UEM string
+        """
+        return "".join([line for line in self._iter_uem()])
+
     def write_uem(self, file: TextIO):
         """Dump timeline to file using UEM format
 
@@ -1081,14 +1107,7 @@ def write_uem(self, file: TextIO):
         >>> with open('file.uem', 'w') as file:
         ...    timeline.write_uem(file)
         """
-
-        uri = self.uri if self.uri else "<NA>"
-        if isinstance(uri, Text) and ' ' in uri:
-            msg = (f'Space-separated UEM file format does not allow file URIs '
-                   f'containing spaces (got: "{uri}").')
-            raise ValueError(msg)
-        for segment in self:
-            line = f"{uri} 1 {segment.start:.3f} {segment.end:.3f}\n"
+        for line in self._iter_uem():
             file.write(line)
 
     def for_json(self):

setup.py

@@ -34,7 +34,7 @@
 
     # package
     namespace_packages=['pyannote'],
-    packages=find_packages(),
+    packages=find_packages(exclude=["tests"]),
     install_requires=[
         'sortedcontainers >= 2.0.4',
         'numpy >= 1.10.4',