@salesforce/afv-skills 1.66.0 → 1.68.0

package/skills/agentforce-offline-experimentation-run/scripts/btlib/agentpatch.py

@@ -0,0 +1,483 @@
+"""The same edit ops in the .agent file: the source-side twin of patch.py.
+
+patch.py scores an edit by applying it to the captured requests. This makes that edit in the
+agent's source, so the version that ships is the one that was scored:
+
+    new_text, report = agentpatch.patch_agent(text, patch.parse_ops(doc))
+
+The targets are patch.py's, in the source:
+  node:<name>          the node's reasoning instruction text blocks
+  system               system.instructions, and the system.instructions override of a node in
+                       scope (how an override renders is not verified on traces)
+  tool:<name>, transition:<name>
+                       a transition binding's description; for an action binding, the
+                       description of its action definition
+  tool:<name>.<param>  that input's description, where the binding offers it (with p = ...)
+
+find / anchor / with / text are matched and written verbatim, as .agent source text (dedented,
+merge fields as written). Only the lines an edit changes are rewritten, at the block's indent
+and in its | style; procedure statements are never touched. A quoted description stays one
+quoted string, in JSON escapes (the compiler decodes \\" \\\\ \\n \\t); set_description adds a
+missing description. The result is parsed back: it must be the same agent, with only the edited
+texts changed, as the parser reads them.
+
+An op is refused (AgentPatchError, its .reason one of REASONS) when the .agent edit could
+differ from the request edit that was scored, or when it cannot be written:
+  no_target   the agent lacks the target, or a node named in nodes
+  unmatched   the find / anchor is in none of the target's texts
+  ambiguous   it occurs more than once without all, counted over all of a node's blocks (each
+              branch of an if is a block); set_description over different descriptions
+  render      the source edit renders otherwise than the same edit on the rendered text: the
+              find matches the rendered text but not the source ({!@actions.X} written as X),
+              or more often in the rendered text than in the source
+  shared      the text is also used where the op does not apply: an action definition bound
+              to another tool, or in a node outside nodes; system.instructions with nodes that
+              leave out a node using it
+  unwritable  the result cannot be written as intended, or does not read back as intended
+              (an empty block, a trailing blank line, a CR, a control character, ...)
+"""
+
+import json
+import re
+from difflib import SequenceMatcher
+
+from .agentsrc import MERGE_FIELD, AgentSourceError, _child, _str, parse_agent, render_text
+from .patch import edit_text
+
+REASONS = ("no_target", "unmatched", "ambiguous", "render", "shared", "unwritable")
+DESCRIPTION_INDENT = 4  # a description added to an entry without children: this much deeper
+
+
+class AgentPatchError(ValueError):
+    def __init__(self, reason, message):
+        super().__init__(message)
+        self.reason = reason
+
+
+def _err(op, reason, msg):
+    return AgentPatchError(reason, f"op {op.id}: {msg}")
+
+
+class _Text:
+    """One editable text: a block, a one-line value, or a missing description (text None)."""
+
+    def __init__(self, where, block=None, item=None, owner=None):
+        self.where, self.block, self.item, self.owner = where, block, item, owner
+        if block is not None:
+            self.text = block.text
+        elif item is not None:
+            self.text = _str(item.value)
+        else:
+            self.text = None
+
+
+def _description(where, owner):
+    it = _child(owner, "description")
+    if it is None:
+        return _Text(where, owner=owner)
+    if it.block is not None:
+        return _Text(where, block=it.block)
+    return _Text(where, item=it)
+
+
+def _editable(src):
+    """{key: _Text} for every text an op can edit, and the descriptions of action bindings."""
+    out = {}
+    if src.system is not None:
+        out[("system", None)] = _Text("system.instructions", block=src.system)
+    for n, node in src.nodes.items():
+        if node.system is not None:
+            out[("system", n)] = _Text(f"the system.instructions of node {n}", block=node.system)
+        for k, b in enumerate(node.blocks):
+            out[("block", n, k)] = _Text(f"instruction block {k + 1} of node {n} (line {b.start})", block=b)
+        for b in node.bindings.values():
+            out[("binding", n, b.name)] = _description(f"the description of {b.name} in node {n}", b.item)
+        for d in node.action_defs.values():
+            out[("action", n, d.name)] = _description(f"the description of action {d.name} (node {n})", d.item)
+            for inp in d.inputs.values():
+                out[("input", n, d.name, inp.name)] = _description(
+                    f"the description of input {inp.name} of action {d.name} (node {n})", inp.item)
+    return out
+
+
+def _shape(src):
+    """All of the agent but its editable texts and line numbers: an edit must keep it."""
+    nodes = []
+    for node in src.nodes.values():
+        nodes.append((node.kind, node.name, node.label, node.description, node.model, node.procedure,
+                      node.system is not None, [(b.style, b.conditional) for b in node.blocks],
+                      [(b.name, b.kind, b.target, b.llm_inputs, sorted(b.bound_inputs.items()), b.available_when)
+                       for b in node.bindings.values()],
+                      [(d.name, d.label, d.target,
+                        [(i.name, i.type, i.label, i.is_required, i.is_user_input) for i in d.inputs.values()])
+                       for d in node.action_defs.values()]))
+    return src.model, src.developer_name, src.system is not None, nodes
+
+
+# ---------------------------------------------------------------------------------------
+# targets
+
+def _home(src, d):
+    for n, node in src.nodes.items():
+        if node.action_defs.get(d.name) is d:
+            return n
+    return None
+
+
+def _scope(src, op):
+    if op.nodes is None:
+        return list(src.nodes)
+    missing = sorted(op.nodes - set(src.nodes))
+    if missing:
+        raise _err(op, "no_target", f"nodes {missing} are not in the agent")
+    return [n for n in src.nodes if n in op.nodes]
+
+
+def _targets(src, texts, op):
+    """[(key, _Text)] of the texts the op applies to."""
+    if op.kind == "node":
+        node = src.nodes.get(op.name)
+        if node is None:
+            raise _err(op, "no_target", f"the agent has no node {op.name}")
+        if not node.blocks:
+            raise _err(op, "no_target", f"node {op.name} has no instruction text")
+        keys = [("block", op.name, k) for k in range(len(node.blocks))]
+    elif op.kind == "system":
+        keys = _system_keys(src, op, _scope(src, op))
+    else:
+        keys = _description_keys(src, op, _scope(src, op))
+    return [(k, texts[k]) for k in keys]
+
+
+def _system_keys(src, op, scope):
+    users = [n for n, node in src.nodes.items() if node.system is None]  # nodes on the global text
+    keys = []
+    if src.system is not None and (op.nodes is None or any(n in op.nodes for n in users)):
+        left = [n for n in users if op.nodes is not None and n not in op.nodes]
+        if left:
+            raise _err(op, "shared", f"system.instructions is also the system text of nodes {left}, which nodes "
+                                     "leaves out: list them too, or drop nodes")
+        keys.append(("system", None))
+    keys += [("system", n) for n in scope if src.nodes[n].system is not None]
+    if not keys:
+        raise _err(op, "no_target", "no system.instructions in "
+                   + ("the agent" if op.nodes is None else f"nodes {sorted(op.nodes)}"))
+    return keys
+
+
+def _description_keys(src, op, scope):
+    keys, defs = [], []
+    for n in scope:
+        node = src.nodes[n]
+        b = node.bindings.get(op.name)
+        if b is None:
+            continue
+        if b.kind == "transition":
+            if op.param:
+                raise _err(op, "no_target", f"{op.name} in node {n} is a transition: it has no parameters")
+            key = ("binding", n, op.name)
+        elif b.kind == "action":
+            d = src.action_def(node, b.action)
+            if d is None:
+                raise _err(op, "no_target", f"{op.name} in node {n} is bound to @actions.{b.action}, "
+                                            "which the agent does not define")
+            if op.param:
+                if op.param not in d.inputs:
+                    raise _err(op, "no_target", f"action {d.name} has no input {op.param}")
+                if op.param not in b.llm_inputs:
+                    continue  # bound to an expression: not a parameter of the tool
+                key = ("input", _home(src, d), d.name, op.param)
+            else:
+                key = ("action", _home(src, d), d.name)
+            if d not in defs:
+                defs.append(d)
+        else:
+            raise _err(op, "no_target", f"{op.name} in node {n} is neither an action nor a transition")
+        if key not in keys:
+            keys.append(key)
+    if not keys:
+        where = "the agent" if op.nodes is None else f"nodes {sorted(op.nodes)}"
+        offers = f" with {op.param} = ..." if op.param else ""
+        raise _err(op, "no_target", f"no node of {where} binds {op.name}{offers}")
+    for d in defs:
+        _check_shared_def(src, op, d)
+    return keys
+
+
+def _check_shared_def(src, op, d):
+    """An action definition's text is the description of every tool bound to it: each of
+    those must be the op's tool, in its scope, or the .agent edit changes more tools than the
+    request edit did."""
+    others = []
+    for n, node in src.nodes.items():
+        for b in node.bindings.values():
+            if b.kind != "action" or src.action_def(node, b.action) is not d:
+                continue
+            if op.param and op.param not in b.llm_inputs:
+                continue
+            if b.name != op.name or (op.nodes is not None and n not in op.nodes):
+                others.append(f"{b.name} in node {n}")
+    if others:
+        what = f"input {op.param} of action {d.name}" if op.param else f"action {d.name}"
+        raise _err(op, "shared", f"the description of {what} is also that of {', '.join(others)}, which the op "
+                                 "leaves alone: widen nodes, or give the tool its own action definition first")
+
+
+# ---------------------------------------------------------------------------------------
+# the edits: {key: new text}
+
+def _what(op):
+    word = "find" if op.op in ("replace", "delete") else "anchor"
+    loc = op.locator if len(op.locator) <= 60 else op.locator[:57] + "..."
+    return f"the {word} {loc!r}"
+
+
+def _where(targets):
+    if len(targets) == 1:
+        return targets[0][1].where
+    return f"{len(targets)} texts ({'; '.join(t.where for _, t in targets)})"
+
+
+def _verbatim(text, op, pattern):
+    out, pos = [], 0
+    for m in pattern.finditer(text):
+        if op.op == "insert_after":
+            out += [text[pos:m.end()], op.new]
+        elif op.op == "insert_before":
+            out += [text[pos:m.start()], op.new, m.group(0)]
+        else:
+            out += [text[pos:m.start()], op.new]
+        pos = m.end()
+    out.append(text[pos:])
+    return "".join(out)
+
+
+def _cut_field(text, pattern):
+    """The first merge field that a match of pattern starts or ends inside of, if any."""
+    fields = [(m.start(), m.end(), m.group(0)) for m in MERGE_FIELD.finditer(text)]
+    for m in pattern.finditer(text):
+        for start, end, field in fields:
+            if start < m.end() and m.start() < end and not (m.start() <= start and end <= m.end()):
+                return field
+    return None
+
+
+def _block_edits(op, targets):
+    """The verbatim source edit of each block, checked against the request edit of its
+    rendered text (patch.py's, which the replay scored)."""
+    pattern = re.compile(re.escape(op.locator))
+    for _, t in targets:
+        field = _cut_field(t.text, pattern)
+        if field:
+            raise _err(op, "render", f"{_what(op)} cuts into the merge field {field} in {t.where}: the edit "
+                                     "would change what it refers to. Find whole merge fields, as written")
+    found = {key: len(pattern.findall(t.text)) for key, t in targets}
+    for key, t in targets:
+        if not found[key] and edit_text(render_text(t.text), op)[1] != "unmatched":
+            raise _err(op, "render", f"{_what(op)} matches the rendered text of {t.where} but not its source: copy "
+                                     "it from the .agent, with merge fields ({!@actions.X}) and the indentation of "
+                                     "its lines as written")
+    total = sum(found.values())
+    if not total:
+        raise _err(op, "unmatched", f"{_what(op)} is not in {_where(targets)}: copy it from the .agent")
+    hit = [key for key, _ in targets if found[key]]
+    if total > 1 and not op.all:
+        branches = " (in different blocks: each branch of an if is one)" if len(hit) > 1 else ""
+        raise _err(op, "ambiguous", f"{_what(op)} occurs {total} times in {_where(targets)}{branches}: "
+                                    "lengthen it, or set all")
+    edits = {}
+    for key, t in targets:
+        if not found[key]:
+            continue
+        new = _verbatim(t.text, op, pattern)
+        req_new, outcome = edit_text(render_text(t.text), op)
+        if outcome == "ambiguous":
+            raise _err(op, "render", f"{_what(op)} occurs once in the source of {t.where} but more often in its "
+                                     "rendered text ({!@actions.X} renders as X): lengthen it")
+        if outcome != "matched" or render_text(new) != req_new:
+            raise _err(op, "render", f"the edit of {t.where} renders otherwise than the same edit on its rendered "
+                                     "text (a line that starts with a merge field loses its indent): keep merge "
+                                     "fields off the start of a line the edit changes")
+        edits[key] = new
+    return edits
+
+
+def _description_edits(op, targets):
+    if op.op == "set_description":
+        current = {t.text for _, t in targets}
+        if len(current) > 1:
+            raise _err(op, "ambiguous", f"{op.target} has {len(current)} different descriptions in scope "
+                                        f"({'; '.join(t.where for _, t in targets)}): limit the op with nodes")
+        return {key: op.new for key, _ in targets}
+    edits, matched = {}, False
+    for key, t in targets:
+        new, outcome = edit_text(t.text or "", op)
+        if outcome == "ambiguous":
+            raise _err(op, "ambiguous", f"{_what(op)} occurs more than once in {t.where}: lengthen it, or set all")
+        if outcome == "matched":
+            matched = True
+            edits[key] = new
+    if not matched:
+        raise _err(op, "unmatched", f"{_what(op)} is not in {_where(targets)}: copy it from the .agent")
+    return edits
+
+
+# ---------------------------------------------------------------------------------------
+# writing: (first line index, end index, new raw lines) splices of the file's lines
+
+def _block_raws(op, t, new):
+    """The block's new raw lines: unchanged lines as they were, changed ones at its indent."""
+    b = t.block
+    if not b.lines:
+        raise _err(op, "unwritable", f"{t.where} has no text lines")
+    old, lines = b.text.split("\n"), new.split("\n")
+    if not new.strip():
+        raise _err(op, "unwritable", f"the edit would leave {t.where} empty")
+    if not lines[-1].strip():
+        raise _err(op, "unwritable", f"the edit would end {t.where} with a blank line, which is not part of a block "
+                                     "in the .agent: take the line break before the text into the find instead")
+    if any(ln and not ln.strip() for ln in lines):
+        raise _err(op, "unwritable", f"the edit would give {t.where} a line of only spaces")
+    inline = b.style == "inline"
+    if inline and (not lines[0] or lines[0].startswith(" ")):
+        raise _err(op, "unwritable", f"the edit would leave the | line of {t.where} empty or starting with a space")
+    if not inline and next(ln for ln in lines if ln).startswith(" "):
+        raise _err(op, "unwritable", f"the edit would indent the first text line of {t.where}, which sets the "
+                                     "block's indent")
+    raws = [raw for _, raw in b.lines]
+
+    def fmt(j, text):
+        if inline and j == 0:
+            return raws[0][:b.base] + text
+        return " " * b.base + text if text else ""
+
+    out = []
+    for tag, i1, _, j1, j2 in SequenceMatcher(None, old, lines, autojunk=False).get_opcodes():
+        for d in range(j2 - j1):
+            i, j = i1 + d, j1 + d
+            keep = tag == "equal" and not (inline and (i == 0) != (j == 0))
+            out.append(raws[i] if keep else fmt(j, lines[j]))
+    return out
+
+
+def _string(op, t, new):
+    bad = sorted({hex(ord(c)) for c in new if ord(c) < 32 and c not in "\n\t"})
+    if bad:
+        raise _err(op, "unwritable", f"the new text of {t.where} holds control characters {bad}")
+    return json.dumps(new, ensure_ascii=False)
+
+
+def _quoted_raw(op, t, new, lines):
+    it = t.item
+    body = lines[it.line - 1].rstrip()
+    try:
+        one_string = isinstance(json.loads(it.value), str)
+    except ValueError:
+        one_string = False
+    if it.children or not one_string or not body.endswith(it.value):
+        raise _err(op, "unwritable", f"{t.where} is not a one-line quoted string: make it one, or a "
+                                     "description: | block, first")
+    return body[:len(body) - len(it.value)] + _string(op, t, new)
+
+
+def _missing_raw(op, t, new):
+    owner = t.owner
+    indent = owner.children[0].indent if owner.children else owner.indent + DESCRIPTION_INDENT
+    return " " * indent + "description: " + _string(op, t, new)
+
+
+_KINDS = {"block": "instructions", "system": "system", "binding": "transition", "action": "action",
+          "input": "input"}
+
+
+def _about(key):
+    out = {"text": _KINDS[key[0]]}
+    if key[0] == "block":
+        out.update(node=key[1], block=key[2])
+    elif key[0] == "system":
+        if key[1] is not None:
+            out["node"] = key[1]
+    else:
+        out.update(node=key[1], name=key[2])
+        if key[0] == "input":
+            out["input"] = key[3]
+    return out
+
+
+def _hunks(key, start, old, new):
+    out = []
+    for tag, i1, i2, j1, j2 in SequenceMatcher(None, old, new, autojunk=False).get_opcodes():
+        if tag != "equal":
+            out.append(dict(_about(key), line=start + i1 + 1, removed=old[i1:i2], added=new[j1:j2]))
+    return out
+
+
+def _apply(text, src, op):
+    """(new text, new source, report edits) of one op."""
+    if "\r" in op.new:
+        raise _err(op, "unwritable", "the op's text holds a carriage return")
+    texts = _editable(src)
+    targets = _targets(src, texts, op)
+    edits = (_description_edits if op.is_description else _block_edits)(op, targets)
+    lines = text.split("\n")
+    splices = []
+    for key, t in targets:
+        new = edits.get(key)
+        if new is None or new == t.text:
+            continue
+        if t.block is not None:
+            splices.append((t.block.start - 1, t.block.end, _block_raws(op, t, new), key))
+        elif t.item is not None:
+            splices.append((t.item.line - 1, t.item.line, [_quoted_raw(op, t, new, lines)], key))
+        else:
+            splices.append((t.owner.line, t.owner.line, [_missing_raw(op, t, new)], key))
+    report = []
+    for i0, i1, raws, key in sorted(splices, key=lambda s: s[0], reverse=True):
+        report[:0] = _hunks(key, i0, lines[i0:i1], raws)
+        lines[i0:i1] = raws
+    new_text = "\n".join(lines)
+    return new_text, _read_back(op, src, texts, new_text, edits), report
+
+
+def _parsed(src, key):
+    """The text at key as the parser reads it, which is what node_tools gives the scorer; it is
+    the one _editable found unless an entry holds two descriptions (the parser keeps the last)."""
+    if key[0] == "system":
+        return (src.system if key[1] is None else src.nodes[key[1]].system).text
+    node = src.nodes[key[1]]
+    if key[0] == "block":
+        return node.blocks[key[2]].text
+    if key[0] == "binding":
+        return node.bindings[key[2]].description
+    d = node.action_defs[key[2]]
+    return d.description if key[0] == "action" else d.inputs[key[3]].description
+
+
+def _read_back(op, src, texts, new_text, edits):
+    """The edited agent, parsed: the same agent with only the edited texts changed."""
+    try:
+        new_src = parse_agent(new_text, src.path)
+    except AgentSourceError as e:
+        raise _err(op, "unwritable", f"the edited agent does not parse: {e}")
+    if _shape(new_src) != _shape(src):
+        raise _err(op, "unwritable", "the edit changes more than texts (nodes, blocks, bindings or definitions "
+                                     "read back otherwise)")
+    wrong = [t.where for key, t in texts.items() if _parsed(new_src, key) != edits.get(key, _parsed(src, key))]
+    if wrong:
+        raise _err(op, "unwritable", f"the edited agent reads back otherwise in {'; '.join(wrong)}")
+    return new_src
+
+
+def patch_agent(text, ops, path=None):
+    """(new text, report) of the ops applied in order to an .agent's source text, each to the
+    result of the ones before; no ops gives the text unchanged. The report has one entry per
+    op, {id, target, op, edits}; an edit is {text, node, block | name [, input], line,
+    removed, added}, its line 1-based in the text the op applied to."""
+    if "\r" in text:
+        raise AgentPatchError("unwritable", f"{path or 'agent'}: has CR line endings; convert it to LF first")
+    src = parse_agent(text, path)
+    report = []
+    for op in ops:
+        text, src, edits = _apply(text, src, op)
+        report.append({"id": op.id, "target": op.target, "op": op.op, "edits": edits})
+    return text, report