[tools] Remove compile-time dependencies on lxml (RobotLocomotion#19027)

Programs invoked at compile-time should only use standard Python modules
for maximum portability.

Luckily in this case, Python 3 ships with a (slightly less featureful)
ElementTree implementation that we can use.

doc/build.py

@@ -9,7 +9,7 @@
 import urllib.parse
 
 from bazel_tools.tools.python.runfiles import runfiles
-import lxml.etree as ET
+import xml.etree.ElementTree as ET
 
 from drake.doc.defs import check_call, main
 
@@ -134,9 +134,9 @@ def _build_sitemap(site_dir: str) -> None:
         loc = ET.SubElement(url, "loc")
         loc.text = location
     sitemap = ET.ElementTree(urlset)
+    ET.indent(sitemap)
     sitemap.write(os.path.join(site_dir, "sitemap.xml"),
                   encoding="utf-8",
-                  pretty_print=True,
                   xml_declaration=True)
 
 

setup/mac/binary_distribution/requirements.txt

@@ -1,5 +1,4 @@
 ipywidgets
-lxml
 matplotlib
 notebook
 Pillow

setup/ubuntu/binary_distribution/packages-focal.txt

@@ -44,7 +44,6 @@ libxt6
 ocl-icd-libopencl1
 python3
 python3-ipywidgets
-python3-lxml
 python3-matplotlib
 python3-numpy
 python3-pil

setup/ubuntu/binary_distribution/packages-jammy.txt

@@ -45,7 +45,6 @@ libxt6
 ocl-icd-libopencl1
 python3
 python3-ipywidgets
-python3-lxml
 python3-matplotlib
 python3-munkres
 python3-numpy

setup/ubuntu/source_distribution/packages-focal-test-only.txt

@@ -5,7 +5,6 @@ llvm-12
 python3-dateutil
 python3-dbg
 python3-flask
-python3-lxml-dbg
 python3-matplotlib-dbg
 python3-nbconvert
 python3-nbformat

tools/wheel/image/provision-python.sh

@@ -41,7 +41,6 @@ ln -s /usr/local/bin/python /usr/bin/python
 # TODO(jwnimmer-tri): Should these be version-pinned? What's the process for
 # keeping them up to date if they are?
 pip install \
-    lxml \
     matplotlib \
     numpy \
     pyyaml \

tools/workspace/pybind11/pybind_coverage_xml_parser.py

@@ -1,15 +1,22 @@
-from lxml import etree as ET
-import pandas
-import os
 import logging
+import os
+import xml.etree.ElementTree as ET
+
+import pandas
 
 
 XPATHS = {
-    "class_decl": ".//Node[@kind='CursorKind.CLASS_DECL' or "
-    "@kind='CursorKind.CLASS_TEMPLATE' or @kind='CursorKind.STRUCT_DECL']"
-    "[@ignore='0']",
-    "file_names": ".//Node",
-    "doc_var_xpath": ".//Node[@ignore='0' and @doc_var]",
+    "class_decl": [
+        ".//Node[@kind='CursorKind.CLASS_DECL'][@ignore='0']",
+        ".//Node[@kind='CursorKind.CLASS_TEMPLATE'][@ignore='0']",
+        ".//Node[@kind='CursorKind.STRUCT_DECL'][@ignore='0']",
+    ],
+    "file_names": [
+        ".//Node",
+    ],
+    "doc_var_xpath": [
+        ".//Node[@ignore='0'][@doc_var]"
+    ],
 }
 
 all_kinds = [
@@ -162,7 +169,9 @@ def write_coverage(self):
         """Writes class-wise coverage to a CSV file.
 
         """
-        class_nodes = self.xml_root.xpath(XPATHS["class_decl"])
+        class_nodes = []
+        for xpath in XPATHS["class_decl"]:
+            class_nodes.extend(self.xml_root.findall(xpath))
         self.df = pandas.DataFrame(
                 columns=["ClassName", "Coverage"] + all_kinds)
         self.get_class_coverage(class_nodes)
@@ -181,7 +190,9 @@ def get_class_coverage(self, class_nodes):
 
         for c in class_nodes:
             # All nodes which are not ignored and have a doc_var
-            doc_var_nodes = c.xpath(XPATHS["doc_var_xpath"])
+            doc_var_nodes = []
+            for xpath in XPATHS["doc_var_xpath"]:
+                doc_var_nodes.extend(c.findall(xpath))
             row = CommonUtils.get_node_coverage(
                     doc_var_nodes, self.pybind_strings)
 
@@ -191,6 +202,7 @@ def get_class_coverage(self, class_nodes):
 
         self.df_pruned = CommonUtils.prune_dataframe(
                 self.df, ["ClassName", "Coverage"])
+        self.df_pruned = self.df_pruned.sort_values(by=["ClassName"])
 
 
 class FileCoverage:
@@ -216,7 +228,9 @@ def write_coverage(self):
         """Writes file-wise coverage to a CSV file.
 
         """
-        file_nodes = self.xml_root.xpath(XPATHS["doc_var_xpath"])
+        file_nodes = []
+        for xpath in XPATHS["doc_var_xpath"]:
+            file_nodes.extend(self.xml_root.findall(xpath))
         self.df = pandas.DataFrame(
                 columns=["DirCoverage", "FileName", "Coverage"] + all_kinds)
 
@@ -300,7 +314,7 @@ def add_directory_coverage(self):
         filenames, coverage = df.loc[:, "FileName"], df.loc[:, "Coverage"]
         file_coverage = dict(zip(filenames, coverage))
         root = self.make_tree(file_coverage)
-        XP = root.xpath
+        XP = root.findall
         dirname = None
         final_row = {}
 
@@ -312,10 +326,12 @@ def add_directory_coverage(self):
             if dirname != os.path.dirname(str(row["FileName"])):
                 dirname = os.path.dirname(str(row["FileName"]))
                 total_num = sum([
-                        int(n) for n in XP('.//{}/*/@num'.format(dirname))
+                        int(n.get("num"))
+                        for n in XP('.//{}/*/[@num]'.format(dirname))
                     ])
                 total_den = sum([
-                        int(d) for d in XP('.//{}/*/@den'.format(dirname))
+                        int(d.get("den"))
+                        for d in XP('.//{}/*/[@den]'.format(dirname))
                     ])
             row["DirCoverage"] = Coverage(total_num, total_den)
 

tools/workspace/sdformat_internal/embed_sdf.py

@@ -2,7 +2,7 @@
 
 import sys
 
-from lxml import etree
+import xml.etree.ElementTree as ET
 
 assert __name__ == '__main__'
 
@@ -11,16 +11,17 @@ def _minified_xml(*, filename):
     """Given a filename for an `*.sdf` schema file, returns a minified xml
     string with its contents, to conserve disk space in Drake's library.
     """
-    parser = etree.XMLParser(remove_blank_text=True, remove_comments=True)
-    root = etree.parse(filename, parser)
-    for item in root.xpath("//description"):
-        item.getparent().remove(item)
-    for elem in root.iter('*'):
+    tree = ET.parse(filename)
+    # Remove all '<description>' elements.
+    for item in tree.findall(".//description/.."):
+        item.remove(item.find("description"))
+    # Discard whitespace.
+    for elem in tree.iter('*'):
         if elem.text is not None:
             elem.text = elem.text.strip()
         if elem.tail is not None:
             elem.tail = elem.tail.strip()
-    return etree.tostring(root, encoding="utf-8", xml_declaration=False)
+    return ET.tostring(tree.getroot(), encoding="utf-8", xml_declaration=False)
 
 
 filenames = sorted(sys.argv[1:])