KiPdZddlmZddlZddlZddlZddlZddlZddlm Z ddl m Z ddl m Z mZddlZddlmZmZmZdd lmZmZdd lmZmZej6eZd Ze Gd d Ze GddZ e GddZ!GddZ"dddZ#ddZ$ddZ%y)aSQLite-backed knowledge graph storage and query engine. Stores code structure as nodes (File, Class, Function, Type, Test) and edges (CALLS, IMPORTS_FROM, INHERITS, IMPLEMENTS, CONTAINS, TESTED_BY, DEPENDS_ON, REFERENCES). Supports impact-radius queries and subgraph extraction. ) annotationsN) dataclass)Path)AnyOptional) BFS_ENGINEMAX_IMPACT_DEPTHMAX_IMPACT_NODES)get_schema_versionrun_migrations)EdgeInfoNodeInfoar CREATE TABLE IF NOT EXISTS nodes ( id INTEGER PRIMARY KEY AUTOINCREMENT, kind TEXT NOT NULL, -- File, Class, Function, Type, Test name TEXT NOT NULL, qualified_name TEXT NOT NULL UNIQUE, file_path TEXT NOT NULL, line_start INTEGER, line_end INTEGER, language TEXT, parent_name TEXT, params TEXT, return_type TEXT, modifiers TEXT, is_test INTEGER DEFAULT 0, file_hash TEXT, extra TEXT DEFAULT '{}', updated_at REAL NOT NULL ); CREATE TABLE IF NOT EXISTS edges ( id INTEGER PRIMARY KEY AUTOINCREMENT, kind TEXT NOT NULL, -- CALLS, IMPORTS_FROM, INHERITS, REFERENCES, etc. source_qualified TEXT NOT NULL, target_qualified TEXT NOT NULL, file_path TEXT NOT NULL, line INTEGER DEFAULT 0, extra TEXT DEFAULT '{}', confidence REAL DEFAULT 1.0, confidence_tier TEXT DEFAULT 'EXTRACTED', updated_at REAL NOT NULL ); CREATE TABLE IF NOT EXISTS metadata ( key TEXT PRIMARY KEY, value TEXT NOT NULL ); CREATE INDEX IF NOT EXISTS idx_nodes_file ON nodes(file_path); CREATE INDEX IF NOT EXISTS idx_nodes_kind ON nodes(kind); CREATE INDEX IF NOT EXISTS idx_nodes_qualified ON nodes(qualified_name); CREATE INDEX IF NOT EXISTS idx_edges_source ON edges(source_qualified); CREATE INDEX IF NOT EXISTS idx_edges_target ON edges(target_qualified); CREATE INDEX IF NOT EXISTS idx_edges_kind ON edges(kind); CREATE INDEX IF NOT EXISTS idx_edges_target_kind ON edges(target_qualified, kind); CREATE INDEX IF NOT EXISTS idx_edges_source_kind ON edges(source_qualified, kind); CREATE INDEX IF NOT EXISTS idx_edges_file ON edges(file_path); ceZdZUded<ded<ded<ded<ded<ded<ded <ded <d ed <d ed <d ed<ded<d ed<ded<y) GraphNodeintidstrkindnamequalified_name file_path line_startline_endlanguage Optional[str] parent_nameparams return_typeboolis_test file_hashdictextraN__name__ __module__ __qualname____annotations__c/home/jay/workspace/scripts/.codegraph-venv/lib/python3.12/site-packages/code_review_graph/graph.pyrrQsM G I INOMM  M Kr+rcpeZdZUded<ded<ded<ded<ded<ded<d ed <d Zd ed <dZded<y) GraphEdgerrrrsource_qualifiedtarget_qualifiedrliner#r$?float confidence EXTRACTEDconfidence_tierN)r&r'r(r)r4r6r*r+r,r.r.cs; G IN I KJ&OS&r+r.cTeZdZUded<ded<ded<ded<ded<ded <d ed <y ) GraphStatsr total_nodes total_edgeszdict[str, int] nodes_by_kind edges_by_kind list[str] languages files_countr last_updatedNr%r*r+r,r8r8ps,!!!!r+r8ceZdZdZd=dZd>dZd?dZd?dZd?dZd?dZ d@dAdZ dBd Z dCd Z d@ dDd Z dEd ZdFd ZdGdZd?dZd?dZdHdZdIdZdJdKdZdLdZdLdZdMdNdZ dO dPdZdQdZdRdZdSdTdZeef dUdZeef dUdZ eef dUdZ!dVdZ"dWdZ# dX dYd!Z$dZd"Z% d[ d\d#Z&d]d$Z'd^d%Z(d_d&Z) d`d'Z*dad(Z+dbd)Z, dcd*Z-ddd+Z. ded,Z/ dfd-Z0dgd.Z1dhd/Z2did0Z3 dbd1Z4 djd2Z5 dkd3Z6 dld4Z7 dmd5Z8dnd6Z9dod7Z:dpd8Z;dqd9Zdtd<Z?y )u GraphStorez#SQLite-backed code knowledge graph.ct||_|jjjddt j t |jddd|_tj|j_ |jjd|jjd|jt|jdkr5|jjd |jjt|jd|_t!j"|_y) NT)parentsexist_okF)timeoutcheck_same_threadisolation_levelzPRAGMA journal_mode=WALzPRAGMA busy_timeout=5000rzJINSERT OR IGNORE INTO metadata (key, value) VALUES ('schema_version', '1'))rdb_pathparentmkdirsqlite3connectr_connRow row_factoryexecute _init_schemar commitr _nxg_cache threadingLock _cache_lock)selfrJs r,__init__zGraphStore.__init__sG}  !!$!>__  rU   ")  45 56  djj )A - JJ  1  JJ   tzz"-1$>>+r+c|SNr*rYs r, __enter__zGraphStore.__enter__s r+c$|jyr\)close)rYexc_typeexc_valexc_tbs r,__exit__zGraphStore.__exit__s  r+cv|jjt|jjyr\)rO executescript _SCHEMA_SQLrTr]s r,rSzGraphStore._init_schemas$   - r+cT|j5d|_dddy#1swYyxYw)z>,<= jnn-> LM :&::%% 3YY T[[$..$)) L  (*   JJ  +E:Xd^T  D> !  5YY T[[$..$))U # /  zz!!">?HHJ1MMr+c|jjd|f|jjd|f|jy)z2Remove all nodes and edges associated with a file.z%DELETE FROM nodes WHERE file_path = ?z%DELETE FROM edges WHERE file_path = ?N)rOrRri)rYrs r,remove_file_datazGraphStore.remove_file_datas< BYLQ BYLQ  r+c|jjr/tjd|jj |jj d |j ||D]}|j|||D]}|j||jj |jy#t$r|jjwxYw)z'Atomically replace all data for a file.z;Flushing unexpected open transaction before BEGIN IMMEDIATEBEGIN IMMEDIATEr"N) rOin_transactionloggerwarningrTrRrrvr~ BaseExceptionrollbackri)rYrnodesedgesfhashrrr{s r,store_file_nodes_edgesz!GraphStore.store_file_nodes_edgess :: $ $ NNX Y JJ    ,-   ! !) , 8   7 8 '  & ' JJ       JJ   !  s "AC%C5c~|jjd |D]J\}}}}|j||D]}|j|||D]}|j |L|jj |jy#t $r|jjwxYw)z@Atomically replace data for a batch of files in one transaction.rrN) rOrRrrvr~rTrrri)rYbatchrrrrrrr{s r,store_file_batchzGraphStore.store_file_batchs ,- 27 +. 5%%%i0!O (* /33!!!$333rc|jjd|fj}|Dcgc]}|j|c}Scc}w)Nz.SELECT * FROM edges WHERE target_qualified = ?rrs r,get_edges_by_targetzGraphStore.get_edges_by_targetSrrc|jjd||fj}|Dcgc]}|j|c}Scc}w)aSearch for edges where target_qualified matches an unqualified name. CALLS edges often store unqualified target names (e.g. ``generateTestCode``) rather than fully qualified ones (``file.ts::generateTestCode``). This method finds those edges by exact match on the plain function name so that reverse call tracing (callers_of) works even when qualified-name lookup returns nothing. z;SELECT * FROM edges WHERE target_qualified = ? AND kind = ?r)rYrrrrs r,search_edges_by_target_namez&GraphStore.search_edges_by_target_nameYsOzz!! I 4L  (* /33!!!$333sA cp|jt}g}|g}jd|fj}|rB|ddk(r:jd|fj D]}|j |ddfd }|D]a} jd| fj D];}|d} | |vs |j | || d } | s+|j | =cd |vr|jd d d n|} jd| fj D];}|d} | |vs |j | || d } | s+|j | =t|} t|D]}t}| D]<} jd| fj D]}|j |d>|D]a}jd|fj D];}|d} | |vs |j | || d } | s+|j | =c|} |S)ajFind tests covering a node, including indirect (transitive) coverage. 1. Direct: TESTED_BY edges targeting this node (+ bare-name fallback). 2. Indirect: follow outgoing CALLS edges up to *max_depth* hops, then collect TESTED_BY edges on each callee. Returns a list of dicts with node fields plus ``indirect: bool``. z/SELECT kind FROM nodes WHERE qualified_name = ?rClasszSSELECT target_qualified FROM edges WHERE source_qualified = ? AND kind = 'CONTAINS'r0ctjd|fj}|sy|d|d|d|d|dS)Nrrrrr)rrrrindirect)rRrq)qnrruconns r, _node_dictz3GraphStore.get_transitive_tests.._node_dictsV,,>hj F "%&6"7 -F $  r+zTSELECT source_qualified FROM edges WHERE target_qualified = ? AND kind = 'TESTED_BY'r/F)r::rzPSELECT target_qualified FROM edges WHERE source_qualified = ? AND kind = 'CALLS'T)rrrr returnz dict | None) rOsetrRrqrappendaddrsplitrange)rYr max_depthseenresults input_qnsrumrowrrsrcdbarefrontier_ next_frontiercalleers @r,get_transitive_testszGraphStore.get_transitive_testshszz $$ ll =    (*  3v;') C!hj  ;   &8!9:  ;  *B||Dhj * ,-d?HHSM"37Aq) * *6:^5K~$$T1-b1Q_<< @ G  (* &C ()C$ sU3NN1% &y>y! %A&)eM ?<<DE(* ?C "%%c*<&=> ? ?( .<<HI(* .C 01C$ &sT:#NN1- . .%H+ %.r+c|j}|jdj}|syi}|jdjD])}|j|dgj |d+i}|jdjD]N}|d}d|vr|j dd dn|}|j|d t j|Pd}|D]} | d} |j| g} | st| d k(r| d} nx| d } d| vr| j dd dn| d }|j|t }| Dcgc]}|j dd d|vr|}}t|d k(r|d} n|jd | | d f|d z }|r&|jtjd||Scc}w)a:Batch-resolve bare-name CALLS targets using the global node table. After parsing, some CALLS edges have bare targets (no ``::`` separator) because the parser couldn't resolve cross-file. This method matches them against nodes and updates unambiguous matches in-place. Disambiguation strategy: 1. Single node with that name -> resolve directly 2. Multiple candidates -> prefer one whose file is imported by the source file (via IMPORTS_FROM edges) Returns the number of resolved edges. z}SELECT id, source_qualified, target_qualified, file_path FROM edges WHERE kind = 'CALLS' AND target_qualified NOT LIKE '%::%'rzRSELECT name, qualified_name FROM nodes WHERE kind IN ('Function', 'Test', 'Class')rrzRSELECT DISTINCT file_path, target_qualified FROM edges WHERE kind = 'IMPORTS_FROM'r0rrrr/z2UPDATE edges SET target_qualified = ? WHERE id = ?rz#Resolved %d bare-name CALLS targets) rOrRr setdefaultrsplitrrrxlenrTrinfo)rYr bare_edges node_lookupruimport_targetsrz target_fileresolvedr{ bare_name candidatesrtsrc_qnsrc_fileimported_filescimporteds r,resolve_bare_call_targetsz$GraphStore.resolve_bare_call_targetss7zz\\ S  (* -/ << :  (* RC  " "3v; 3 : :3?O;P Q  R/1<< *  (* PC+,F6:fn&,,tQ/2&K  % %c+&6 > B B; O  P D/0I$B7J:!#&qM 0104FLLq)!,k*"0!3!3Hce!D)wwtQ'*n<x=A% ( I LLDDJ'  MH; >  KKM KK=x H%s$ G#c|jjdj}|Dcgc]}|d c}Scc}w)Nz8SELECT DISTINCT file_path FROM nodes WHERE kind = 'File'rrOrRrrYrrs r, get_all_fileszGraphStore.get_all_filess=zz!! F (* )--1+--- =c|j}|sgS t|dk(rd|jddzdz}ndjd|D}|jj d||fj }|r|Dcgc]}|j|c}S g}g}|D]>} | j} |jd|jd| dd| dg@dj|} d | d } |j||jj | |j }|Dcgc]}|j|c}Scc}w#t$rYwxYwcc}w) zKeyword search across node names. Tries FTS5 first (fast, tokenized matching), then falls back to LIKE-based substring search when FTS5 returns no results. r""" AND c3LK|]}d|jddzdzyw)rrN)replace).0ws r, z*GraphStore.search_nodes..-s*)9:C!))C..4)s"$zZSELECT n.* FROM nodes_fts f JOIN nodes n ON f.rowid = n.id WHERE nodes_fts MATCH ? LIMIT ?z4(LOWER(name) LIKE ? OR LOWER(qualified_name) LIKE ?)%SELECT * FROM nodes WHERE z LIMIT ?) rrrjoinrOrRrr Exceptionlowerrextend) rYquerylimitwords fts_queryrr conditionsrwordrwheresqls r, search_nodeszGraphStore.search_nodess  I 5zQ%--T"::S@ #LL)>C) ::%%2E"  hj  6:;))!,;; !# "$ 0D A   F  MMQqc8q1X. /  0 Z(*5': ezz!!#v.779.23!!!$33%<   "4s*A1E EE E/E E,+E,cdtdk(r|j|||S|j|||S)aBFS from changed files to find all impacted nodes within depth N. Delegates to ``get_impact_radius_sql()`` by default (faster for large graphs). Set ``CRG_BFS_ENGINE=networkx`` to use the legacy Python-side BFS via NetworkX. Returns dict with: - changed_nodes: nodes in changed files - impacted_nodes: nodes reachable via edges - impacted_files: unique set of affected files - edges: connecting edges networkx)r max_nodes)r _get_impact_radius_networkxget_impact_radius_sql)rY changed_filesrrs r,get_impact_radiuszGraphStore.get_impact_radiusMsL$  #33i4 )) Y)*  r+c |s ggggdddSt}|D]5}|j|}|D]}|j|j7|s ggggdddS|jj d|jj dd}t |} tdt| |D]A} | | | |z} djd| D} |jj d | | Cd } |jj | |||t|zfj}t}|D]}|d}||vs |j||j|}|j|}t|}||kD}|r|d |}t |Dchc]}|jc}}g}||Dchc]}|jc}z}|r|j|}||||||dScc}wcc}w) zImpact radius via SQLite recursive CTE. Faster than NetworkX for large graphs because it avoids materialising the full graph in Python. Fr changed_nodesimpacted_nodesimpacted_filesr truncatedtotal_impactedzCCREATE TEMP TABLE IF NOT EXISTS _impact_seeds (qn TEXT PRIMARY KEY)zDELETE FROM _impact_seeds,c3 K|]}dyw)z(?)Nr*rrs r,rz3GraphStore.get_impact_radius_sql..s#9aE#9 z0INSERT OR IGNORE INTO _impact_seeds (qn) VALUES a` WITH RECURSIVE impacted(node_qn, depth) AS ( SELECT qn, 0 FROM _impact_seeds UNION SELECT e.target_qualified, i.depth + 1 FROM impacted i JOIN edges e ON e.source_qualified = i.node_qn WHERE i.depth < ? UNION SELECT e.source_qualified, i.depth + 1 FROM impacted i JOIN edges e ON e.target_qualified = i.node_qn WHERE i.depth < ? ) SELECT DISTINCT node_qn, MIN(depth) AS min_depth FROM impacted GROUP BY node_qn LIMIT ? N)rrrrrOrRlistrrrr_batch_get_nodesrget_edges_among)rYrrrseedsfrn batch_size seed_listir placeholderscte_sqlr impacted_qnsrrrrr r rrelevant_edgesall_qnss r,rz GraphStore.get_impact_radius_sqlisU!#"$"$""#  % ,A**1-E , !**+ , , !#"$"$""#    $  67 K q#i.*5 AaJ/E88#95#99L JJ  B<.Q   &zz!! iIE ,BC (* "%  %A1B  $ % --e4 ..|<^,"Y. +JY7NNCqq{{CD*,^D1++DD !11':N+,,#",   DEs !G6G;c|j}t}|D]5}|j|}|D]}|j|j7t} |j } d} t} | r| |kr| j | t} | D]}||vr=|j|D])}|| vs| j|| j|+||vsI|j|D])}|| vs| j|| j|+| | z} t| t| z|kDrn| } | dz } | r| |kr|j|}| |z }|j|}t|}||kD}|r|d|}t|Dchc]}|jc}}g}||Dchc]}|jc}z}|r|j|}||||||dScc}wcc}w)z=BFS via NetworkX (legacy). Used when CRG_BFS_ENGINE=networkx.rrNr)_build_networkx_graphrrrrcopyupdate neighbors predecessorsrrrrr)rYrrrnxgrrrrvisitedrdepthimpactedrrneighborpredrrrr r rrrs r,rz&GraphStore._get_impact_radius_networkxs/((*% ,A**1-E , !**+ , ,  E::< U59, NN8 $&)eM /9$'MM"$53#72)--h7$LL239 # 0 0 4/w.)--d3$LL./ / W $M7|c-009<$H QJE%59,(--e4 %' ..|<^,"Y. +JY7NNCqq{{CD*,^D1++DD !11':N+,,#",   DEs G2>G7cg}|D]'}|j|}|s|j|)g}t|}|D]8}|j|D]"}|j|vs|j|$:||dS)zMExtract a subgraph containing the specified nodes and their connecting edges.)rr)rrrrr0)rYqualified_namesrrrrrqn_setes r, get_subgraphzGraphStore.get_subgraphs! #B==$D T" # _%! $B--b1 $%%/LLO $ $ //r+c d|jjdjd}|jjdjd}i}|jjdD] }|d||d<i}|jjdD] }|d||d<|jjdDcgc]}|d  }}|jjd jd}|jd } t ||||||| Scc}w) z,Return aggregate statistics about the graph.zSELECT COUNT(*) FROM nodesrzSELECT COUNT(*) FROM edgesz5SELECT kind, COUNT(*) as cnt FROM nodes GROUP BY kindcntrz5SELECT kind, COUNT(*) as cnt FROM edges GROUP BY kindzQSELECT DISTINCT language FROM nodes WHERE language IS NOT NULL AND language != ''rz.SELECT COUNT(*) FROM nodes WHERE kind = 'File'r@)r9r:r;r<r>r?r@)rOrRrqrr8) rYr9r:r;rur<rr>r?r@s r, get_statszGraphStore.get_stats+sJjj(()EFOOQRST jj(()EFOOQRST (* ::%%&]^ 4C),UM#f+ & 4)+ ::%%&]^ 4C),UM#f+ & 4$(::#5#5c$ AjM  jj(( < (*Q ((8 ##''#%   s D-Ncgd}|g}|"|jd|j||r"|jd|j||r&|jd|jd|d|j|dj|}|jjd|d|j } | D cgc]} |j | c} Scc} w) aFind nodes within a line-count range, ordered largest first. Args: min_lines: Minimum line count threshold (inclusive). max_lines: Maximum line count threshold (inclusive). None = no upper bound. kind: Filter by node kind (Function, Class, File, etc.). file_path_pattern: SQL LIKE pattern to filter by file path. limit: Maximum results to return. Returns: List of GraphNode objects, ordered by line count descending. )zline_start IS NOT NULLzline_end IS NOT NULLz (line_end - line_start + 1) >= ?z (line_end - line_start + 1) <= ?zkind = ?file_path LIKE ?rrrz2 ORDER BY (line_end - line_start + 1) DESC LIMIT ?)rrrOrRrr) rY min_lines max_linesrfile_path_patternrrrrrrs r,get_nodes_by_sizezGraphStore.get_nodes_by_sizeNs( "{    @ A MM) $    j ) MM$     0 1 MMA/02 3 e Z(zz!!(0@ @   (* /33!!!$333s C'c|jjd|fj}|r|j|SdS)z/Fetch a single node by its integer primary key.z SELECT * FROM nodes WHERE id = ?NrrYnode_idrus r,get_node_by_idzGraphStore.get_node_by_id~sAjj  .  (* *-t  %6$6r+cr|sgSdjd|D}d|dg}t|}|r&|jd|jd|ddj|}|jj d||j }|Dcgc]}|j |c}Scc}w) aMReturn nodes matching any of *kinds*, optionally filtered by file. Args: kinds: List of node kind strings (e.g. ``["Function", "Test"]``). file_pattern: If provided, only nodes whose ``file_path`` contains *file_pattern* (SQL LIKE ``%pattern%``) are returned. r c3 K|]}dyw?Nr*rs r,rz/GraphStore.get_nodes_by_kind..s33rz kind IN ()r3rrr)rrrrOrRrr) rYkinds file_patternrrrrrrs r,get_nodes_by_kindzGraphStore.get_nodes_by_kindsIxx3U33 !,q12  K    0 1 MMAl^1- . Z(zz!!( 0& (* /33!!!$333sB4cj|jjd|fj}|r|dSdS)z2Return the number of flows a node participates in.z>SELECT COUNT(*) as cnt FROM flow_memberships WHERE node_id = ?r0rrr9s r,count_flow_membershipsz!GraphStore.count_flow_membershipss?jj   J  (* !s5z'a'r+c|jjd|fj}|Dcgc]}|d c}Scc}w)z?Return criticality values for all flows a node participates in.zdSELECT f.criticality FROM flows f JOIN flow_memberships fm ON fm.flow_id = f.id WHERE fm.node_id = ? criticalityr)rYr:rrs r,get_flow_criticalities_for_nodez*GraphStore.get_flow_criticalities_for_nodesIzz!! #J  (* +//Q- /// ?cr|jjd|fj}|r |d|dSy)z4Return the ``community_id`` for a node, or ``None``.z+SELECT community_id FROM nodes WHERE id = ? community_idNrr9s r,get_node_community_idz GraphStore.get_node_community_idsGjj  9 J  (*  3~&2~& &r+ci}d}tdt||D]b}||||z}djd|D}|jj d|d|j }|D] }|d||d<d|S) zBatch-fetch ``community_id`` for a list of qualified names. Returns a mapping from qualified name to community_id (may be ``None`` if the node has no assigned community). r rr c3 K|]}dywr>r*rs r,rzBGraphStore.get_community_ids_by_qualified_names..#7AC#7rzHSELECT qualified_name, community_id FROM nodes WHERE qualified_name IN (r@rKr)rrrrOrRr) rYqnsresultrrrrrrs r,$get_community_ids_by_qualified_namesz/GraphStore.get_community_ids_by_qualified_namess)+ q#c(J/ @A!j.)E88#7#77L::%%,,8><hj   @./.?q)*+ @ @ r+c|jjdd|fj}|Dcgc]}|d c}Scc}w)zr*rs r,rz3GraphStore.get_node_ids_by_files..rOrz)SELECT id FROM nodes WHERE file_path IN (r@c3&K|] }|d yw)rNr*rrs r,rz3GraphStore.get_node_ids_by_files..s0a!D'0)rrrrrOrRrr!)rY file_pathsrQrrrrrs r,get_node_ids_by_filesz GraphStore.get_node_ids_by_filess5L5 q#j/:6 1AqZ0E88#7#77L::%%''3nA7hj  MM040 0 1 r+cf|sgSt|}g}d}tdt||D]h}||||z}djd|D}|jj d|d|j }|jd|Djttj|S)z8Return distinct flow IDs that contain any of *node_ids*.r rr c3 K|]}dywr>r*rs r,rz6GraphStore.get_flow_ids_by_node_ids..rOrz@SELECT DISTINCT flow_id FROM flow_memberships WHERE node_id IN (r@c3&K|] }|d yw)flow_idNr*rds r,rz6GraphStore.get_flow_ids_by_node_ids.."s51!I,5re) rrrrrOrRrrr#fromkeys) rYnode_idsnidsrQrrrrrs r,get_flow_ids_by_node_idsz#GraphStore.get_flow_ids_by_node_idssIH~ q#d)Z0 6A1z>*E88#7#77L::%%%%1N!5hj  MM55 5 6DMM&)**r+c|jjd|fj}|Dchc]}|d c}Scc}w)z6Return the set of qualified names for nodes in a flow.zgSELECT n.qualified_name FROM flow_memberships fm JOIN nodes n ON fm.node_id = n.id WHERE fm.flow_id = ?rr)rYrkrrs r,get_flow_qualified_namesz#GraphStore.get_flow_qualified_names&sIzz!! E J  (* .22"#222rIcj|jjd|fj}|r|dSdS)z8Return just the ``kind`` column for a node, or ``None``.z#SELECT kind FROM nodes WHERE id = ?rNrr9s r,get_node_kind_by_idzGraphStore.get_node_kind_by_id/s:jj  1G: (* "s6{+t+r+c|jjdj}|Dchc]}|d c}Scc}w)z8Return the set of all CALLS-edge target qualified names.z@SELECT DISTINCT target_qualified FROM edges WHERE kind = 'CALLS'r0rrs r,get_all_call_targetszGraphStore.get_all_call_targets6sCzz!! #  (* 044!$%444rc |jjdjS#tj$r"}t j d|gcYd}~Sd}~wwxYw)z/Return raw rows from the ``communities`` table.z SELECT id, name FROM communitiesz0Communities list unavailable (table missing): %sNr\)rYr_s r,get_communities_listzGraphStore.get_communities_list>sT ::%%2hj ''  LLKS QI s(+A AA A c|jjd|fj}|Dcgc]}|d c}Scc}w)z/Return qualified names of nodes in a community.z7SELECT qualified_name FROM nodes WHERE community_id = ?rrrYrKrrs r,get_community_member_qnsz#GraphStore.get_community_member_qnsKsJzz!! % O  (* .22"#222rIc|jjd|fj}|Dcgc]}|j|c}Scc}w)z*Return all nodes belonging to a community.z*SELECT * FROM nodes WHERE community_id = ?rrys r,get_nodes_by_community_idz$GraphStore.get_nodes_by_community_idVsMzz!! 8 O  (* /33!!!$333rcg}d}tdt||D]h}||||z}djd|D}|jj d|d|j }|j d|Dj|S)z@Return ``target_qualified`` for edges sourced from *source_qns*.r rr c3 K|]}dywr>r*rs r,rz2GraphStore.get_outgoing_targets..hrOrz>SELECT target_qualified FROM edges WHERE source_qualified IN (r@c3&K|] }|d yw)r0Nr*rds r,rz2GraphStore.get_outgoing_targets..n?Q1/0?rerrrrOrRrr)rY source_qnsrrrrrrs r,get_outgoing_targetszGraphStore.get_outgoing_targets`  q#j/:6 @AqZ0E88#7#77L::%%..:^1>hj  NN?$? ? @r+cg}d}tdt||D]h}||||z}djd|D}|jj d|d|j }|j d|Dj|S)z=Return ``source_qualified`` for edges targeting *target_qns*.r rr c3 K|]}dywr>r*rs r,rz2GraphStore.get_incoming_sources..yrOrz>SELECT source_qualified FROM edges WHERE target_qualified IN (r@c3&K|] }|d yw)r/Nr*rds r,rz2GraphStore.get_incoming_sources..rrer)rY target_qnsrrrrrrs r,get_incoming_sourceszGraphStore.get_incoming_sourcesqrr+c|jjdj}|Dcgc]}|j|c}Scc}w)zReturn all edges in the graph.SELECT * FROM edgesrrs r, get_all_edgeszGraphStore.get_all_edgess?zz!!"78AAC.23!!!$333sA cn|sgSt|}g}d}tdt||D]}||||z}djd|D}|jj d|d|j }|D]3} |j| } | j|vs#|j| 5|S)zReturn edges where both source and target are in the given set. Batches the source-side IN clause to stay under SQLite's default SQLITE_MAX_VARIABLE_NUMBER limit, then filters targets in Python. r rr c3 K|]}dywr>r*rs r,rz-GraphStore.get_edges_among..rOrz/SELECT * FROM edges WHERE source_qualified IN (r@) rrrrrOrRrrr0r) rYr+rPrrrrrrrr{s r,rzGraphStore.get_edges_amongs I?##% q#c(J/ )A!j.)E88#7#77L::%%A,qQhj  )((+((O;NN4( ) )r+c4|sgSt|}g}d}tdt||D]j}||||z}djd|D}jj d|d|j }|jfd|Dl|S)zJBatch-fetch nodes by qualified name, staying under SQLite variable limits.r rr c3 K|]}dywr>r*rs r,rz.GraphStore._batch_get_nodes..rOrz-SELECT * FROM nodes WHERE qualified_name IN (r@c3@K|]}j|ywr\)r)rrrYs r,rz.GraphStore._batch_get_nodes..s>A4,,Q/>s)rrrrrOrRrr) rYr+rPrrrrrrs ` r,rzGraphStore._batch_get_nodessI?##% q#c(J/ ?A!j.)E88#7#77L::%%? ~QOhj  NN>> > ?r+c\|j5|j|jcdddStj}|jj dj }|D]}|j|d|d|d!||_|cdddS#1swYyxYw)zJBuild (or return cached) in-memory NetworkX directed graph from all edges.Nrr/r0r)r)rXrUnxDiGraphrOrRradd_edge)rYgrrs r,rz GraphStore._build_networkx_graphs    *  JJLA::%%&;<EEGD Y 1/0!4F2GaPVi X YDO   sB"A)B""B+c|jdk(r |jS|jr(|jd|jd|jS|jd|jS)NFiler.)rrrr)rYrrs r,rmzGraphStore._make_qualifiedsb 99 >> !   nn%R(8(8'9499+F F..!DII;//r+ct|d|d|d|d|d|d|d|dxsd |d |d |d t|d |d|drtj|dSiS)Nrrrrrrrrrrrr!r"r$)rrrrrrrrrrrr!r"r$)rr rnloads)rYrus r,rzGraphStore._row_to_nodes4yVV/0+&<(__*M*x=M*Y(+&.1'l$**S\*  AC  r+c |drtj|dni}d|jvr|dnd}d|jvr|dnd}t|d|d|d|d |d |d ||| S) Nr$r4r2r6r5rrr/r0rr1) rrr/r0rr1r$r4r6)rnrkeysr.)rYrur$r4r6s r,rzGraphStore._row_to_edges,/L 3w<(b*6#((**DS&# 4E4S#/0Yd4yV !34 !34+&V!+  r+)rJz str | PathrNone)rz 'GraphStore')rr)r)rrrr"rrr)r{rrr)rrrr) rrrzlist[NodeInfo]rzlist[EdgeInfo]rrrr)rz5list[tuple[str, list[NodeInfo], list[EdgeInfo], str]]rr)rrrrrr)rrrr)rrrOptional[GraphNode])rrrlist[GraphNode])T)rr rr)rrrlist[GraphEdge])CALLS)rrrrrr)r)rrrrrz list[dict])rr)rr=))rrrrrr)rr=rrrrrdict[str, Any])r+r=rr)rr8)2NNNr) r4rr5 int | Noner str | Noner6rrrrr)r:rrrr\)rAr=rBrrr)r:rrr)r:rrz list[float])r:rrr)rPr=rdict[str, int | None])rTrrr=)rzlist[sqlite3.Row])r:rrYrrr)rr)rfr=rset[int])rmrrz list[int])rkrrset[str])r:rrr)rr)rKrrr=)rKrrr)rr=rr=)rr=rr=)rr)r+rrr)r+rrr)rz nx.DiGraph)rrrrr)ru sqlite3.Rowrr)rurrr.)@r&r'r(__doc__rZr^rdrSrir`rvr~rrrrrrTrrrrrrrrrrrr r rrrr.r1r7r;rCrErHrLrRrUrWrZr`rgrorqrsrurwrzr|rrrrrrrmrrr*r+r,rBrBs-,.#  D ND!Z\!!%3!*) l  l l  l  l f*) ?  ? ?  ?  ? B0"! J $(, ,4,4,4 ,4 & ,4  ,4 ,4`7$(44!4  46(0 ,.  '*  $# &+ + +*3,5    3 3  344 4# "# &4 0& 0 $ r+rBc<djd|D}|d|S)aStrip ASCII control characters and truncate to prevent prompt injection. Node names extracted from source code could contain adversarial strings (e.g. ``IGNORE_ALL_PREVIOUS_INSTRUCTIONS``). This function removes control characters (0x00-0x1F except tab and newline) and enforces a length limit so that names flowing through MCP tool responses cannot easily influence AI agent behaviour. rc3DK|]}|dvst|dk\r|yw))  N)ord)rchs r,rz!_sanitize_name..s*  RD s N)r)smax_lencleaneds r,_sanitize_namers.ggG 8G r+c J|j|jt|jt|j|j |j |j|j|jrt|jn |j|jd S)N) rrrrrrrrrr!) rrrrrrrrrrr!)rs r, node_to_dictrskddAFFN166,B()9)9:ll JJ89 ~amm41==99  r+c |j|jt|jt|j|j |j |j|jdS)N)rrryrzrr1r4r6) rrrr/r0rr1r4r6)r-s r, edge_to_dictrsQddAFF !3!34 !3!34[[!&&llq7H7H  r+))rrrrrr)rrrr#)r-r.rr#)&r __future__rrnloggingrMrVrl dataclassesrpathlibrtypingrrrr constantsr r r migrationsr r parserrr getLoggerr&rrgrr.r8rBrrrr*r+r,rs#  ! EE:&   8 $ / d  "  ' '  '     g g T#"r+