rpm_dep_tree: graph of (Build)Requires for a rpm

Author: ydirson

scripts/rpmwatcher/repoquery.py

@@ -0,0 +1,146 @@
+import logging
+import os
+import re
+import subprocess
+from typing import Iterable, Sequence
+
+XCPNG_YUMREPO_TMPL = """
+[xcpng-{section}{suffix}]
+name=xcpng - {section}{suffix}
+baseurl=https://updates.xcp-ng.org/8/{version}/{section}/{rpmarch}/
+gpgkey=https://xcp-ng.org/RPM-GPG-KEY-xcpng
+failovermethod=priority
+skip_if_unavailable=False
+"""
+
+def setup_xcpng_yum_repos(*, yum_repo_d: str, sections: Iterable[str],
+                          bin_arch: str | None, version: str) -> None:
+    with open(os.path.join(yum_repo_d, "xcpng.repo"), "w") as yumrepoconf:
+        for section in sections:
+            # binaries
+            block = XCPNG_YUMREPO_TMPL.format(rpmarch=bin_arch,
+                                              section=section,
+                                              version=version,
+                                              suffix='',
+                                              )
+            yumrepoconf.write(block)
+            # sources
+            block = XCPNG_YUMREPO_TMPL.format(rpmarch='Source',
+                                              section=section,
+                                              version=version,
+                                              suffix='-src',
+                                              )
+            yumrepoconf.write(block)
+
+DNF_BASE_CMD = None
+def dnf_setup(*, dnf_conf: str, yum_repo_d: str) -> None:
+    global DNF_BASE_CMD
+    DNF_BASE_CMD = ['dnf', '--quiet',
+                    '--releasever', 'WTF',
+                    '--config', dnf_conf,
+                    f'--setopt=reposdir={yum_repo_d}',
+                    ]
+
+BINRPM_SOURCE_CACHE: dict[str, str] = {}
+def rpm_source_package(rpmname: str) -> str:
+    return BINRPM_SOURCE_CACHE[rpmname]
+
+def run_repoquery(args: list[str]) -> Sequence[str]:
+    assert DNF_BASE_CMD is not None
+    cmd = DNF_BASE_CMD + ['repoquery'] + args
+    logging.debug('$ %s', ' '.join(cmd))
+    ret = subprocess.check_output(cmd, universal_newlines=True).strip().split()
+    logging.debug('> %s', ret)
+    return ret
+
+SRPM_BINRPMS_CACHE: dict[str, set[str]] = {}         # binrpm-nevr -> srpm-nevr
+def fill_srpm_binrpms_cache() -> None:
+    # HACK: get nevr for what dnf outputs as %{sourcerpm}
+    logging.debug("get epoch info for SRPMs")
+    args = [
+        '--disablerepo=*', '--enablerepo=*-src', '*',
+        '--qf', '%{name}-%{version}-%{release}.src.rpm,%{name}-%{evr}',
+        '--latest-limit=1',
+    ]
+    SRPM_NEVR_CACHE = {         # sourcerpm -> srpm-nevr
+        sourcerpm: nevr
+        for sourcerpm, nevr in (line.split(',')
+                             for line in run_repoquery(args))
+    }
+
+    # binary -> source mapping
+    logging.debug("get binary to source mapping")
+    global SRPM_BINRPMS_CACHE, BINRPM_SOURCE_CACHE
+    args = [
+        '--disablerepo=*-src', '*',
+        '--qf', '%{name}-%{evr},%{sourcerpm}', # FIXME no epoch in sourcerpm, why does it work?
+        '--latest-limit=1',
+    ]
+    BINRPM_SOURCE_CACHE = {
+        # packages without source are not in SRPM_NEVR_CACHE, fallback to sourcerpm
+        binrpm: SRPM_NEVR_CACHE.get(sourcerpm, srpm_strip_src_rpm(sourcerpm))
+        for binrpm, sourcerpm in (line.split(',')
+                                  for line in run_repoquery(args))
+    }
+
+    # reverse mapping source -> binaries
+    SRPM_BINRPMS_CACHE = {}
+    for binrpm, srpm in BINRPM_SOURCE_CACHE.items():
+        binrpms = SRPM_BINRPMS_CACHE.get(srpm, set())
+        if not binrpms:
+            SRPM_BINRPMS_CACHE[srpm] = binrpms
+        binrpms.add(binrpm)
+
+def srpm_nevr(rpmname: str) -> str:
+    args = [
+        '--disablerepo=*', '--enablerepo=*-src',
+        '--qf=%{name}-%{evr}', # to get the epoch only when non-zero
+        '--latest-limit=1',
+        rpmname,
+    ]
+    ret = run_repoquery(args)
+    assert ret, f"Found no SRPM named {rpmname}"
+    assert len(ret) == 1        # ensured by --latest-limit=1 ?
+    return ret[0]
+
+# dnf insists on spitting .src.rpm names it cannot take as input itself
+def srpm_strip_src_rpm(srpmname: str) -> str:
+    SUFFIX = ".src.rpm"
+    assert srpmname.endswith(SUFFIX), f"{srpmname} does not end in .src.rpm"
+    nrv = srpmname[:-len(SUFFIX)]
+    return nrv
+ 
+def rpm_requires(rpmname: str) -> Sequence[str]:
+    args = [
+        '--disablerepo=*-src', # else requires of same-name SRPM are included
+        '--qf=%{name}-%{evr}', # to avoid getting the arch and explicit zero epoch
+        '--resolve',
+        '--requires', rpmname,
+    ]
+    ret = run_repoquery(args)
+    return ret
+
+def srpm_requires(srpmname: str) -> set[str]:
+    args = [
+        '--qf=%{name}-%{evr}', # to avoid getting the arch
+        '--resolve',
+        '--requires', f"{srpmname}.src",
+    ]
+    ret = set(run_repoquery(args))
+    return ret
+
+def srpm_binrpms(srpmname: str) -> set[str]:
+    ret = SRPM_BINRPMS_CACHE.get(srpmname, None)
+    if ret is None: # FIXME should not happen
+        logging.error("%r not found in cache", srpmname)
+        assert False
+        return []
+    logging.debug("binrpms for %s: %s", srpmname, ret)
+    return ret
+
+UPSTREAM_REGEX = re.compile(r'\.el[0-9]+(_[0-9]+)?(\..*|)$')
+RPM_NVR_SPLIT_REGEX = re.compile(r'^(.+)-([^-]+)-([^-]+)$')
+def is_pristine_upstream(rpmname:str) -> bool:
+    if re.search(UPSTREAM_REGEX, rpmname):
+        return True
+    return False

scripts/rpmwatcher/rpm_dep_tree

@@ -0,0 +1,100 @@
+#! /usr/bin/env python3
+
+import atexit
+import logging
+import re
+import sys
+import tempfile
+
+import repoquery
+
+ARCH = "x86_64"
+SHOW_BOUNDARY = False
+
+# Tell if package is pristine upstream, or part of well-kown list of
+# packages we want to consider as "upstream" rather than forks
+def is_upstream(rpmname: str) -> bool:
+    if repoquery.is_pristine_upstream(rpmname):
+        return True
+    m = re.match(repoquery.RPM_NVR_SPLIT_REGEX, rpmname)
+    assert m, f"{rpmname!r} does not match {repoquery.RPM_NVR_SPLIT_REGEX!r}"
+    if m.group(1) in ['systemd', 'util-linux', 'ncurses',
+                      #'xapi',
+                      'devtoolset-11-gcc', 'devtoolset-11-binutils']:
+        return True
+    return False
+
+def main() -> int:
+    logging.basicConfig(format='[%(levelname)s] %(message)s', level=logging.DEBUG)
+
+    this_exe, version, root_srpm = sys.argv
+
+    with (tempfile.NamedTemporaryFile() as dnfconf,
+          tempfile.TemporaryDirectory() as yumrepod,
+          open(f"{root_srpm}-{version}.dot", "w") as dotfile):
+
+        repoquery.setup_xcpng_yum_repos(yum_repo_d=yumrepod,
+                                        sections=['base', 'updates'],
+                                        bin_arch=ARCH,
+                                        version=version)
+        repoquery.dnf_setup(dnf_conf=dnfconf.name, yum_repo_d=yumrepod)
+
+        repoquery.fill_srpm_binrpms_cache()
+        # print([x for x in sorted(SRPM_BINRPMS_CACHE.keys()) if x.startswith("openssl-")])
+        # return 0
+
+        print("digraph packages {", file=dotfile)
+        srpms_seen: set[str] = set()
+        new_srpms = {repoquery.srpm_nevr(root_srpm)}
+        while new_srpms:
+            next_srpms = set()  # preparing next round's new_srpms
+            logging.info("seen: %s, new: %s", len(srpms_seen), len(new_srpms))
+            logging.debug(" new: %s", new_srpms)
+            for srpm in new_srpms:
+                # draw source packages themselves
+                if is_upstream(srpm):
+                    if SHOW_BOUNDARY:
+                        print(f'"{srpm}" [color=grey];', file=dotfile)
+                    logging.debug("skipping upstream %s", srpm)
+                    continue # we don't rebuild upstream rpms
+                elif ".xcpng8.3.":
+                    print(f'"{srpm}";', file=dotfile)
+                else:
+                    print(f'"{srpm}" [color=red];', file=dotfile)
+
+                # build reqs
+                breqs = {repoquery.rpm_source_package(breq)
+                         for breq in repoquery.srpm_requires(srpm)}
+                logging.debug("%s req sources: %s", len(breqs), breqs)
+
+                # reqs of binary rpms produced
+                reqs = set()
+                for binrpm in repoquery.srpm_binrpms(srpm):
+                    reqs.update({repoquery.rpm_source_package(req)
+                                 for req in repoquery.rpm_requires(binrpm)})
+
+                # draw breqs, plain
+                for breq in breqs:
+                    if (not SHOW_BOUNDARY) and is_upstream(breq):
+                        continue
+                    print(f'"{srpm}" -> "{breq}";', file=dotfile)
+                # draw additional runtime reqs, dotted
+                for req in reqs.difference(breqs):
+                    if (not SHOW_BOUNDARY) and is_upstream(req):
+                        continue
+                    if srpm == req:
+                        continue # dependency between RPMs of this SRPM
+                    print(f'"{srpm}" -> "{req}" [style=dotted];', file=dotfile)
+
+                # accumulate
+                srpms_seen.update(new_srpms)
+                next_srpms.update(breqs.difference(srpms_seen))
+                next_srpms.update(reqs.difference(srpms_seen))
+
+            new_srpms = next_srpms
+
+        print("}", file=dotfile)
+        return 0
+
+if __name__ == "__main__":
+    sys.exit(main())