@aryacode/dsh-engineering 0.0.0-stage → 0.3.0

package/vendor/mwi/code/engine/motionworks_iec_mcp/project.py

@@ -0,0 +1,704 @@
+"""Resolve, inventory, and read MotionWorks IEC projects on disk.
+
+A MotionWorks project on disk is an *expanded directory* plus a small sibling
+``.mwt`` file.  In the sample projects the ``.mwt`` is only 4096 bytes: it is a
+wrapper, not the project.  Everything editable lives in the expanded directory::
+
+    <Project>/
+        src.st1                 project tree + view state (CFB)
+        LIST.POU                textual POU list
+        NODES.LST               node/task assignment list
+        @LIBRARY.LST            library references
+        PROJECT.INF             last-change stamp
+        NODES.LST ...
+        C/                      configuration (holds VAR_GLOBAL + IO config)
+        DT/  HW/  LIB/          data types, hardware, library nodes
+        POE/<POU>/              one directory per POU
+            src.st1             <POU>T.TXT, <POU>V.VGR, <POU>V.VB, <POU>.STB|.GB
+            NodeProperties.xml  worksheet GUIDs
+            <POU>.CCI           POU descriptor
+            *.xml               translation documents
+
+Two layout generations appear in the sample data: ``C/Configuration/R/Resource``
+and the older ``C/Resource/R/Resource``.  Both are handled by searching for the
+resource directory rather than assuming a fixed path.
+"""
+
+from __future__ import annotations
+
+import re
+from dataclasses import dataclass, field
+from pathlib import Path
+
+from .cfb import CfbError, CompoundFile
+from .errors import (
+    AmbiguousProject,
+    NotFound,
+    ProjectNotFound,
+    UnsupportedFormat,
+)
+from .variables import (
+    DeclarationTable,
+    Variable,
+    cross_check,
+    decode_declarations,
+    read_grid_variable_count,
+)
+
+#: Extensions that mark a directory as an expanded MotionWorks project.
+_ROOT_MARKERS = ("src.st1", "LIST.POU", "PROJECT.INF")
+
+#: Default directory searched when a caller passes only a project name.
+DEFAULT_SEARCH_ROOT = "Yaskawa"
+
+_GLOBAL_STREAM_RE = re.compile(r"^GLOBAL_VARIABLES", re.IGNORECASE)
+
+
+@dataclass
+class PouInfo:
+    """One POU directory."""
+
+    name: str
+    directory: Path
+
+    @property
+    def source_path(self) -> Path:
+        return self.directory / "src.st1"
+
+    @property
+    def node_properties_path(self) -> Path:
+        return self.directory / "NodeProperties.xml"
+
+    def source(self) -> CompoundFile:
+        return CompoundFile(self.source_path)
+
+    # -- stream discovery -------------------------------------------------
+
+    def stream_names(self) -> list[str]:
+        try:
+            return self.source().stream_names()
+        except CfbError:
+            return []
+
+    def body_stream(self) -> tuple[str, str] | None:
+        """Return ``(stream_name, language)`` for the POU body.
+
+        Graphical streams share the .GB extension. Resolve LD/FBD from the
+        saved worksheet node; the extension alone cannot distinguish them.
+        """
+        streams = self.stream_names()
+        il = [name for name in streams if name.upper().endswith('.AB')]
+        if il:
+            if len(il) != 1 or any(name.upper().endswith(('.STB', '.GB')) for name in streams):
+                raise ValueError('IL body stream is ambiguous')
+            from .tree import parse_document
+            root = self.directory.parent.parent
+            document = parse_document(CompoundFile(root/'src.st1').read_stream('PROJECT.TRE').decode('latin1'))
+            expected = ('POE/'+self.name+'/'+il[0]).casefold()
+            matches = [node for node,_ in document.walk_with_ancestors()
+                       if node.path.split('\t')[0].replace('\\','/').casefold() == expected]
+            if document.warnings or len(matches) != 1 or document.lines[matches[0].start_line].strip() != '9':
+                raise ValueError('IL worksheet tree identity absent or ambiguous')
+            return il[0], 'IL'
+        for name in streams:
+            if name.upper().endswith(".STB"):
+                return name, "ST"
+        for name in streams:
+            if name.upper().endswith(".GB"):
+                from .tree import parse_document
+                root = self.directory.parent.parent
+                document = parse_document(CompoundFile(root / 'src.st1').read_stream('PROJECT.TRE').decode('latin1'))
+                expected = ('POE/' + self.name + '/' + name).casefold()
+                matches = [node for node, _ in document.walk_with_ancestors()
+                           if node.path.split('\t')[0].replace('\\', '/').casefold() == expected]
+                if document.warnings or len(matches) != 1:
+                    raise ValueError('Graphical worksheet tree identity absent or ambiguous')
+                # The leading record kind is 11=LD / 12=FBD. node_id is
+                # the separate, project-specific handle in the params line.
+                record_kind = document.lines[matches[0].start_line].strip()
+                return name, {'11': 'LD', '12': 'FBD'}.get(record_kind, 'GRAPHICAL')
+        return None
+
+    def language(self) -> str | None:
+        found = self.body_stream()
+        return found[1] if found else None
+
+    def declaration_stream_name(self) -> str | None:
+        for name in self.stream_names():
+            if name.upper().endswith("V.VB"):
+                return name
+        return None
+
+    def grid_stream_name(self) -> str | None:
+        for name in self.stream_names():
+            if name.upper().endswith("V.VGR"):
+                return name
+        return None
+
+    def comment_stream_name(self) -> str | None:
+        for name in self.stream_names():
+            if name.upper().endswith("T.TXT"):
+                return name
+        return None
+
+    # -- content ----------------------------------------------------------
+
+    def declarations(self) -> DeclarationTable:
+        cfb = self.source()
+        stream_name = self.declaration_stream_name()
+        if stream_name is None:
+            return DeclarationTable(
+                warnings=["no <POU>V.VB declaration stream present"],
+                source_stream=None,
+            )
+        table = decode_declarations(cfb.read_stream(stream_name), stream_name)
+        grid_name = self.grid_stream_name()
+        if grid_name:
+            table.warnings.extend(
+                cross_check(table, read_grid_variable_count(cfb.read_stream(grid_name)))
+            )
+        return table
+
+    def st_body(self) -> str | None:
+        """Return the Structured Text body, or None for a graphical POU."""
+        found = self.body_stream()
+        if not found or found[1] != "ST":
+            return None
+        return self.source().read_stream(found[0]).decode("latin1")
+
+    def st_body_text(self) -> str | None:
+        """Return readable ST, restoring comments from native translations."""
+        return self.text_body_text() if self.language() == 'ST' else None
+
+    def text_body(self) -> str | None:
+        """Read supported ST/IL source without interpreting graphical bytes."""
+        found = self.body_stream()
+        if not found or found[1] not in ('ST', 'IL'):
+            return None
+        body = self.source().read_stream(found[0]).decode('latin1')
+        if '\0' in body:
+            raise ValueError('Text worksheet contains unsupported binary data')
+        return body
+
+    def text_body_text(self) -> str | None:
+        """Return readable ST/IL with exact native comment translations."""
+        body = self.text_body()
+        if body is None or '\x07' not in body:
+            return body
+        from .st_comments import resolve_comments
+        worksheet = self.body_stream()[0].rsplit('.', 1)[0]
+        matches = [p for p in self.directory.iterdir()
+                   if p.name.casefold() == (worksheet + 'Translation.xml').casefold()]
+        if len(matches) != 1:
+            raise ValueError('Native ST comment translation is absent or ambiguous')
+        return resolve_comments(body, matches[0])
+
+    def summary(self) -> dict[str, object]:
+        try:
+            stream_names = self.stream_names()
+        except CfbError as exc:
+            return {"name": self.name, "error": str(exc)}
+        body = self.body_stream()
+        return {
+            "name": self.name,
+            "language": body[1] if body else None,
+            "body_stream": body[0] if body else None,
+            "streams": len(stream_names),
+            "directory": str(self.directory),
+        }
+
+
+@dataclass
+class Project:
+    """An expanded MotionWorks project directory."""
+
+    root: Path
+    _pous: list[PouInfo] | None = field(default=None, repr=False)
+
+    def __post_init__(self) -> None:
+        # Accept str for convenience: every path operation below assumes a Path,
+        # and a raw string would fail late and confusingly.
+        if not isinstance(self.root, Path):
+            self.root = Path(self.root)
+
+    # -- identity ---------------------------------------------------------
+
+    @property
+    def name(self) -> str:
+        return self.root.name
+
+    @property
+    def mwt_path(self) -> Path | None:
+        candidate = self.root.with_suffix(".mwt")
+        return candidate if candidate.is_file() else None
+
+    @property
+    def node_list_path(self) -> Path | None:
+        for candidate in (self.root / "NODES.LST",):
+            if candidate.is_file():
+                return candidate
+        return None
+
+    # -- raw file reads (ANSI/latin1, as MotionWorks writes them) ---------
+
+    def _read_text(self, name: str) -> str | None:
+        path = self.root / name
+        if not path.is_file():
+            return None
+        return path.read_bytes().decode("latin1")
+
+    def list_pou_lines(self) -> list[str]:
+        text = self._read_text("LIST.POU")
+        return [line for line in (text or "").splitlines() if line.strip()]
+
+    def library_entries(self) -> list[str]:
+        text = self._read_text("@LIBRARY.LST")
+        if not text:
+            return []
+        # First line is a version banner ("Library List, V40").
+        return [line for line in text.splitlines()[1:] if line.strip()]
+
+    def node_lines(self) -> list[str]:
+        text = self._read_text("NODES.LST")
+        return [line for line in (text or "").splitlines() if line.strip()]
+
+    def last_change(self) -> str | None:
+        text = self._read_text("PROJECT.INF")
+        if not text:
+            return None
+        for line in text.splitlines():
+            if line.lower().startswith("lastchange"):
+                return line.split("=", 1)[-1].strip()
+        return None
+
+    def project_tree_streams(self) -> dict[str, int]:
+        path = self.root / "src.st1"
+        if not path.is_file():
+            return {}
+        return CompoundFile(path).summary()
+
+    # -- POUs -------------------------------------------------------------
+
+    @property
+    def pou_directory(self) -> Path:
+        return self.root / "POE"
+
+    def pous(self) -> list[PouInfo]:
+        if self._pous is None:
+            base = self.pou_directory
+            if not base.is_dir():
+                self._pous = []
+            else:
+                self._pous = [
+                    PouInfo(name=child.name, directory=child)
+                    for child in sorted(base.iterdir())
+                    if child.is_dir() and (child / "src.st1").is_file()
+                ]
+        return self._pous
+
+    def pou(self, name: str) -> PouInfo:
+        target = name.casefold()
+        for pou in self.pous():
+            if pou.name.casefold() == target:
+                return pou
+        available = ", ".join(p.name for p in self.pous()) or "<none>"
+        raise NotFound(f"No POU named {name!r} in {self.name}. Available: {available}")
+
+    # -- tasks ------------------------------------------------------------
+
+    def task_assignments(self) -> dict[str, list[str]]:
+        """Map task name -> assigned program names.
+
+        Read from the project tree, which is authoritative.  ``NODES.LST`` is the
+        fallback, and is genuinely worse: ``RotaryKnife_ASP_v350`` has no
+        ``NODES.LST`` at all, so a list-only reader reports no tasks for a project
+        that has four.
+
+        In the tree a task node's path points at the resource task directory and
+        its final component equals its name (``...\\Resource\\FastTsk`` for the
+        node ``FastTsk``); matching the directory alone also catches
+        ``Global_Variables`` and ``IO_Configuration``, whose paths live there but
+        carry a file extension.
+        """
+        from_tree = self._task_assignments_from_tree()
+        if from_tree:
+            return from_tree
+        return self._task_assignments_from_node_list()
+
+    def _task_assignments_from_tree(self) -> dict[str, list[str]]:
+        from .tree import TreeParseError, parse_tree, task_assignments
+
+        path = self.root / "src.st1"
+        if not path.is_file():
+            return {}
+        try:
+            text = CompoundFile(path).read_stream("PROJECT.TRE").decode("latin1")
+            roots, _ = parse_tree(text)
+        except (OSError, ValueError, TreeParseError):
+            return {}
+        return task_assignments(roots)
+
+    def _task_assignments_from_node_list(self) -> dict[str, list[str]]:
+        """Read assignments from ``NODES.LST`` files.
+
+        Loose lines look like::
+
+            TASK     5  MedTsk      eCLR  MP2600iec
+            PROGRAM  6  ServoHoming ServoHoming eCLR MP2600iec
+
+        Both the project root and the resource directory are checked, because
+        projects differ in where they keep the file.  Where both exist they
+        describe the same instances, so programs are de-duplicated in order --
+        without that, every program is reported twice.
+        """
+        assignments: dict[str, list[str]] = {}
+        sources = [self.node_list_path]
+        for extra in (
+            self.root / "C/Configuration/R/Resource/NODES.LST",
+            self.root / "C/Resource/R/Resource/NODES.LST",
+        ):
+            sources.append(extra)
+
+        for source in sources:
+            if source is None or not source.is_file():
+                continue
+            try:
+                lines = source.read_bytes().decode("latin1").splitlines()
+            except OSError:
+                continue
+            current_task: str | None = None
+            for line in lines:
+                if not line.strip():
+                    continue
+                parts = line.split("\t")
+                kind = parts[0].strip().upper()
+                if kind == "TASK" and len(parts) >= 3:
+                    current_task = parts[2].strip()
+                    assignments.setdefault(current_task, [])
+                elif kind == "PROGRAM" and len(parts) >= 3 and current_task:
+                    program = parts[2].strip()
+                    seen = assignments[current_task]
+                    if program not in seen:
+                        seen.append(program)
+        return assignments
+
+    # -- variables --------------------------------------------------------
+
+    def global_variables(self) -> DeclarationTable:
+        """Read ``VAR_GLOBAL`` declarations from the resource source.
+
+        If the resource stores only the binary ``.VGR`` grid (as the newer
+        TopCutter and RotaryKnife projects do), the textual declarations are
+        absent and that is reported rather than invented.
+        """
+        resource = self.resource_source()
+        if resource is None:
+            return DeclarationTable(warnings=["no resource src.st1 found"])
+
+        cfb = CompoundFile(resource)
+        names = cfb.stream_names()
+        # The resource-level stream is named "Global_Variables.VB", which does
+        # not end in "V.VB" (that pattern is for "<POU>V.VB"), so match on the
+        # plain extension here.
+        decl_names = [n for n in names if n.upper().endswith(".VB")]
+        grid_names = [n for n in names if n.upper().endswith(".VGR")]
+
+        if not decl_names:
+            note = "resource src.st1 has no textual declaration stream"
+            if grid_names:
+                count = read_grid_variable_count(cfb.read_stream(grid_names[0]))
+                note += (
+                    f"; the binary grid {grid_names[0]!r} records {count} variables, "
+                    f"but their names and types are not stored as text"
+                )
+            return DeclarationTable(warnings=[note], source_stream=None)
+
+        table = decode_declarations(cfb.read_stream(decl_names[0]), decl_names[0])
+        if grid_names:
+            table.warnings.extend(
+                cross_check(table, read_grid_variable_count(cfb.read_stream(grid_names[0])))
+            )
+        return table
+
+    def resource_source(self) -> Path | None:
+        """Locate the resource ``src.st1`` that holds ``VAR_GLOBAL`` data.
+
+        Handles both observed layouts without hardcoding either one.
+        """
+        for pattern in (
+            "C/Configuration/R/Resource/src.st1",
+            "C/Resource/R/Resource/src.st1",
+        ):
+            candidate = self.root / pattern
+            if candidate.is_file():
+                return candidate
+        # Fall back to a search, preferring a directory named Resource.
+        matches = [
+            p for p in self.root.glob("C/**/src.st1")
+            if p.parent.name.lower() == "resource"
+        ]
+        return matches[0] if matches else None
+
+    # -- reporting --------------------------------------------------------
+
+    def inventory(self) -> dict[str, object]:
+        pous = self.pous()
+        languages: dict[str, int] = {}
+        for pou in pous:
+            languages[pou.language() or "?"] = languages.get(pou.language() or "?", 0) + 1
+        return {
+            "name": self.name,
+            "root": str(self.root),
+            "mwt": str(self.mwt_path) if self.mwt_path else None,
+            "last_change": self.last_change(),
+            "pou_count": len(pous),
+            "languages": languages,
+            "tasks": self.task_assignments(),
+            "libraries": self.library_entries(),
+            "tree_streams": self.project_tree_streams(),
+            "resource_source": str(self.resource_source()) if self.resource_source() else None,
+        }
+
+
+# ---------------------------------------------------------------------------
+# Discovery
+# ---------------------------------------------------------------------------
+
+
+def looks_like_project(directory: Path) -> bool:
+    """True when ``directory`` holds an expanded MotionWorks project."""
+    if not directory.is_dir():
+        return False
+    if (directory / "POE").is_dir():
+        return True
+    return any((directory / marker).is_file() for marker in _ROOT_MARKERS)
+
+
+def discover(search_root: str | Path) -> list[Path]:
+    """Return every expanded project directory under ``search_root``.
+
+    A directory qualifies when it contains a ``POE`` subdirectory (this is the
+    reliable marker: every sample project has one, including ones missing some
+    root metadata files).
+    """
+    base = Path(search_root)
+    if not base.is_absolute():
+        base = Path.cwd() / base
+    if not base.is_dir():
+        return []
+    found = {p.parent for p in base.rglob("POE") if p.is_dir()}
+    return sorted(found)
+
+
+def _match_by_name(base: Path, wanted: str) -> list[Path]:
+    """Match projects by name, supporting a disambiguating path suffix.
+
+    Two of the sample projects are both called ``TopCutter`` (one at the top
+    level, one under ``MP2600iec Program``), so a bare name is genuinely
+    ambiguous.  Callers can qualify it::
+
+        TopCutter                        -> matches both
+        MP2600iec Program/TopCutter      -> matches one
+        Yaskawa/TopCutter                -> matches one
+
+    A supplied multi-segment query must match the project's trailing path
+    components, which is unambiguous without needing an absolute path.
+    """
+    parts = [
+        p.casefold()
+        for p in re.split(r"[\\/]+", wanted.strip())
+        if p and p not in (".", "..")
+    ]
+    if not parts:
+        return []
+    matches: list[Path] = []
+    for candidate in discover(base):
+        tail = [p.casefold() for p in candidate.parts[-len(parts):]]
+        if tail == parts:
+            matches.append(candidate)
+    return matches
+
+
+def resolve(
+    project: str | Path,
+    search_root: str | Path | None = None,
+) -> Project:
+    """Resolve a project by path, by ``.mwt`` file, or by bare/qualified name.
+
+    Accepts an expanded directory, an ``.mwt`` file, or a name that is searched
+    for under ``search_root`` (default :data:`DEFAULT_SEARCH_ROOT`).
+    """
+    candidate = Path(project)
+    if candidate.exists():
+        resolved = candidate.resolve()
+        if resolved.is_file():
+            if resolved.suffix.lower() == ".mwt":
+                expanded = resolved.with_suffix("")
+                if expanded.is_dir():
+                    return Project(root=expanded)
+                raise ProjectNotFound(
+                    f"{resolved.name} has no expanded project directory beside it "
+                    f"({expanded}). Open the project in MotionWorks once to expand it."
+                )
+            raise ProjectNotFound(
+                f"{resolved} is a file; expected an expanded project directory or .mwt"
+            )
+        if looks_like_project(resolved):
+            return Project(root=resolved)
+        raise ProjectNotFound(
+            f"{resolved} does not look like an expanded MotionWorks project "
+            f"(no POE directory or project markers)"
+        )
+
+    # Treat the argument as a name (possibly path-qualified) and search.  Try the
+    # whole query first: a query containing a separator is itself a qualified
+    # suffix such as "MP2600iec Program/TopCutter", and matching on just the last
+    # component would wrongly discard the qualifier.
+    base = Path(search_root) if search_root else Path(DEFAULT_SEARCH_ROOT)
+    if not base.is_absolute():
+        base = Path.cwd() / base
+    raw = str(project)
+    matches = _match_by_name(base, raw)
+    if not matches and candidate.name and candidate.name != raw:
+        matches = _match_by_name(base, candidate.name)
+    if len(matches) == 1:
+        return Project(root=matches[0])
+    if len(matches) > 1:
+        listing = "\n  ".join(str(m) for m in matches)
+        raise AmbiguousProject(
+            f"{project!r} matches {len(matches)} projects; qualify it with a "
+            f"parent directory or pass a full path:\n  {listing}"
+        )
+    available = discover(base)
+    listing = "\n  ".join(str(p) for p in available) or "<none found>"
+    raise ProjectNotFound(
+        f"No project named {project!r} under {base}. Available projects:\n  {listing}"
+    )
+
+
+def describe_all(search_root: str | Path | None = None) -> list[dict[str, object]]:
+    """Inventory every discovered project, for the list tool."""
+    base = Path(search_root) if search_root else Path(DEFAULT_SEARCH_ROOT)
+    if not base.is_absolute():
+        base = Path.cwd() / base
+    out: list[dict[str, object]] = []
+    for path in discover(base):
+        project = Project(root=path)
+        out.append(
+            {
+                "name": project.name,
+                "root": str(path),
+                "pou_count": len(project.pous()),
+                "languages": _language_counts(project),
+                "last_change": project.last_change(),
+            }
+        )
+    return out
+
+
+def _language_counts(project: Project) -> dict[str, int]:
+    counts: dict[str, int] = {}
+    for pou in project.pous():
+        key = pou.language() or "?"
+        counts[key] = counts.get(key, 0) + 1
+    return counts
+
+
+def library_usage(project: Project) -> dict[str, int]:
+    """Count how many POUs declare each function-block type.
+
+    Useful context for an agent: it shows which library blocks the project
+    actually depends on, and therefore which donor POUs are good examples.
+    """
+    counts: dict[str, int] = {}
+    for pou in project.pous():
+        try:
+            table = pou.declarations()
+        except UnsupportedFormat:
+            continue
+        for variable in table.variables:
+            if variable.is_fb_instance:
+                counts[variable.type_name] = counts.get(variable.type_name, 0) + 1
+    return dict(sorted(counts.items(), key=lambda kv: -kv[1]))
+
+
+def symbol_index(project: Project) -> dict[str, list[str]]:
+    """Map every declared symbol name to the POU(s) that declare it.
+
+    This is the cross-reference an agent needs before editing code, so it can
+    avoid inventing symbol names that do not exist.
+    """
+    index: dict[str, list[str]] = {}
+
+    def record(name: str, where: str) -> None:
+        index.setdefault(name, [])
+        if where not in index[name]:
+            index[name].append(where)
+
+    try:
+        for variable in project.global_variables().variables:
+            record(variable.name, "<global>")
+    except UnsupportedFormat:
+        pass
+
+    for pou in project.pous():
+        try:
+            table = pou.declarations()
+        except UnsupportedFormat:
+            continue
+        for variable in table.variables:
+            record(variable.name, pou.name)
+    return index
+
+
+def known_symbol_names(project: Project) -> set[str]:
+    """Every name that could legitimately appear in a POU body.
+
+    This is the allowance list for the ST linter: declared symbols from the
+    whole project plus every type name in use.  It is intentionally generous.
+    Flagging a name that exists elsewhere would be a false positive, and a
+    linter that cries wolf gets ignored; a name that is genuinely undefined will
+    still be caught, because it appears nowhere at all.
+    """
+    names: set[str] = set(symbol_index(project).keys())
+    try:
+        names |= {v.type_name for v in project.global_variables().variables}
+    except UnsupportedFormat:
+        pass
+    for pou in project.pous():
+        try:
+            names |= {v.type_name for v in pou.declarations().variables}
+        except UnsupportedFormat:
+            continue
+    return names
+
+
+def global_names(project: Project) -> set[str]:
+    """The names of the project's VAR_GLOBAL declarations.
+
+    Separate from ``known_symbol_names``, which is deliberately generous, because a
+    global is reachable from a POU ONLY if that POU declares it as VAR_EXTERNAL.  The
+    linter needs to tell the two apart: a name that resolves nowhere is a spelling
+    mistake, while a global this POU has not imported is a build that STALLS with an
+    empty Errors pane.  Measured with PLCMODE_ON, a system global the project has had
+    from the first byte.
+    """
+    try:
+        return {v.name for v in project.global_variables().variables}
+    except UnsupportedFormat:
+        return set()
+
+
+def find_symbol(project: Project, symbol: str) -> list[Variable]:
+    """Return every declaration of ``symbol`` anywhere in the project."""
+    found: list[Variable] = []
+    try:
+        found.extend(project.global_variables().variants(symbol))
+    except UnsupportedFormat:
+        pass
+    for pou in project.pous():
+        try:
+            found.extend(pou.declarations().variants(symbol))
+        except UnsupportedFormat:
+            continue
+    return found