Update benchmarks, add just command & comments

Author: mbollmann

python/benchmarks/bench_attrs.py

@@ -13,71 +13,56 @@
 # limitations under the License.
 
 import attrs
-from attrs import define, field
-
-
-REPEAT = 100_000
-URL = "http://www.lrec-conf.org/proceedings/lrec2000/pdf/374.pdf"
-
-
-class UrlWithoutAttrs:
-    def __init__(self, url: str) -> None:
-        self.url = url
-
-
-@define
-class UrlWithoutValidation:
-    url: str
-
-
-@define
-class UrlWithValidation:
-    url: str = field()
-
-    @url.validator
-    def noop(self, _attribute, _value):
-        pass
-
-
-def vanilla_class_without_validation():
-    """Instantiate a regular class."""
+from pathlib import Path
+
+from acl_anthology.collections import Collection, Volume
+from acl_anthology.people import NameSpecification as NameSpec
+from acl_anthology.text import MarkupText
+
+
+REPEAT = 1_000
+
+
+def create_volume():
+    volume_title = MarkupText.from_string("Lorem ipsum")
+    volume_shorttitle = MarkupText.from_string("L.I.")
+    parent = Collection("2023.acl", None, Path("."))
+    volume = Volume(
+        id="long",
+        parent=parent,
+        type="proceedings",
+        booktitle=volume_title,
+        year="2023",
+        address="Online",
+        doi="10.100/0000",
+        editors=[NameSpec("Bollmann, Marcel")],
+        ingest_date="2023-01-12",
+        isbn="0000-0000-0000",
+        month="jan",
+        pdf=None,
+        publisher="Myself",
+        shortbooktitle=volume_shorttitle,
+        venue_ids=["li", "acl"],
+    )
+
+
+def instantiate_volume_regularly():
+    """Instantiate a Volume."""
     for _ in range(REPEAT):
-        UrlWithoutAttrs(URL)
+        create_volume()
 
 
-def attrs_class_without_validation():
-    """Instantiate a class without attribute validation."""
+def instantiate_volume_without_validation():
+    """Instantiate a class with attribute validation disabled."""
     for _ in range(REPEAT):
-        UrlWithoutValidation(URL)
-
-
-def attrs_class_with_validation():
-    """Instantiate a class with attribute validation."""
-    for _ in range(REPEAT):
-        UrlWithValidation(URL)
-
-
-def attrs_class_with_validation_disabled():
-    """Instantiate a class that has attribute validation, but disabled."""
-    with attrs.validators.disabled():
-        for _ in range(REPEAT):
-            UrlWithValidation(URL)
+        with attrs.validators.disabled():
+            create_volume()
 
 
 __benchmarks__ = [
     (
-        vanilla_class_without_validation,
-        attrs_class_without_validation,
-        "attrs: Vanilla class vs. attrs class",
-    ),
-    (
-        attrs_class_with_validation,
-        attrs_class_without_validation,
-        "attrs: Attribute validation vs. no validation",
-    ),
-    (
-        attrs_class_with_validation_disabled,
-        attrs_class_without_validation,
-        "attrs: Attribute validation disabled vs. no validation",
+        instantiate_volume_regularly,
+        instantiate_volume_without_validation,
+        "attrs: instantiate Volume with attrs.validators.disabled",
     ),
 ]

python/benchmarks/bench_sanitycheck.py

@@ -50,11 +50,11 @@ def detect_url_startswith_protocol():
     (
         detect_url_regex,
         detect_url_contains_separator,
-        "Check URL via regex vs. '://' in string",
+        "Check URL via '://' in string instead of regex",
     ),
     (
         detect_url_startswith_protocol,
         detect_url_contains_separator,
-        "Check URL via .startswith('http') vs. '://' in string",
+        "Check URL via '://' in string instead of .startswith('http')",
     ),
 ]

python/benchmarks/bench_utils.py

@@ -19,7 +19,7 @@
 
 REPEAT = 1_000
 SCRIPTDIR = os.path.dirname(os.path.realpath(__file__))
-XMLFILE = Path(f"{SCRIPTDIR}/../tests/toy_anthology/xml/2022.acl.xml")
+XMLFILE = Path(f"{SCRIPTDIR}/../tests/data/anthology/xml/2022.acl.xml")
 
 # An example from 2022.acl.xml
 tree = etree.parse(XMLFILE)
@@ -38,8 +38,8 @@ def element_tostring():
 
 __benchmarks__ = [
     (
-        element_deepcopy,
         element_tostring,
-        "XML: deepcopy <abstract> vs. convert to string",
+        element_deepcopy,
+        "XML <abstract>: deepcopy instead of etree.tostring",
     ),
 ]

python/benchmarks/bench_xml_markup.py

@@ -87,8 +87,8 @@ def bench_with_iter():
 
 __benchmarks__ = [
     (
-        bench_with_iter,
         bench_with_findtext,
-        "XML: parse <author> via iteration vs. findtext",
+        bench_with_iter,
+        "XML <author>: iterate instead of searching with .findtext",
     ),
 ]

python/benchmarks/bench_xml_names.py

@@ -19,7 +19,7 @@
 
 REPEAT = 3
 SCRIPTDIR = os.path.dirname(os.path.realpath(__file__))
-XMLFILE = Path(f"{SCRIPTDIR}/../tests/toy_anthology/xml/2022.acl.xml")
+XMLFILE = Path(f"{SCRIPTDIR}/../tests/data/anthology/xml/2022.acl.xml")
 
 # Names of XML elements that may appear multiple times, and should be accumulated as a list
 LIST_ELEMENTS = (
@@ -181,13 +181,13 @@ def bench_with_parse_and_clear_element():
 
 __benchmarks__ = [
     (
-        bench_with_parse_element,
         bench_with_parse_single_element,
-        "XML: parse entire <paper> vs. parse one child per function call",
+        bench_with_parse_element,
+        "XML <paper>: one function call instead of one per child tag",
     ),
     (
         bench_with_parse_element,
         bench_with_parse_and_clear_element,
-        "XML: parse <paper> without vs. with clearing them afterwards",
+        "XML <paper>: clear <paper> element after parsing",
     ),
 ]

python/benchmarks/bench_xml_parsing.py

@@ -70,6 +70,14 @@ test-only-xfail: _deps
 typecheck: _deps
   poetry run mypy acl_anthology
 
+# Run benchmarks
+benchmark: _deps
+  poetry run richbench --percentage benchmarks/
+  @echo ""
+  @echo "  Note: Benchmark descriptions should be interpreted as 'Is it faster to...' or"
+  @echo "        'What is the performance impact if we...', with the highlighted columns"
+  @echo "        showing the delta."
+
 # Build the documentation
 docs: _deps
   poetry run mkdocs build