`gjܕdZddlZddlmZmZmZmZddlmZddddddd d d Z d e d e fdZ d>de de de de d ee e ee ee ff dZde deee e fde de d ee f dZde d e fdZd e d ee fdZde de de d e fdZde de deee e fd e fdZde deee e fde de d e f dZ d?de deee e fde dee d e f dZde de d eee e ffdZde de d eee e ffd Zde de d eee e ffd!Zde de d eee e ffd"Zde de d eee e ffd#Zde de d eee e ffd$Zd%e d ee fd&Zd'ee d(eee e fd eee e ffd)Zde de d eee e ffd*Zde de d eee e ffd+Zde de d eee e ffd,Z d-ee d.e d/e d0e d ee e ff d1Z!de d-ee d2ee de d3e d eee e ff d4Z"d%e d5e d6eee e fd eee e ffd7Z#d@de de d8e d9e d e f d:Z$d;ee dz'_detect_escape_drift..KZUCgeC0K)r>r?r&zNEscape-drift detected: old_string and new_string contain the literal sequence a but the matched region of the file does not. This is almost always a tool-call serialization artifact where an apostrophe or quote got prefixed with a spurious backslash. Re-read the file with read_file and pass old_string/new_string without backslash-escaping z characters.)join)rr8rrmatched_regionssuspectplains` rr1r1s  J5 #: ggK7KKO!  j W %:wo?]AJE((/{3& ',Yl <   rlinecpd}|t|kr"||dvr|dz }|t|kr||dvr|d|S)z=Return the leading whitespace prefix of a line (spaces/tabs).r)r  r&Nr0)rOis r_leading_whitespacerTsI A c$i-DG{2 Q c$i-DG{2 8OrcX|jdD]}|js|cSy)zReturn the first line of ``text`` that has any non-whitespace content. Returns ``None`` if no such line exists (text is empty or all whitespace).  N)splitstrip)r rOs r_first_meaningful_linerYs/  4  ::<K r file_regionc|s|St|}t|}|||St|}t|}||k(r|Sg}|jdD]}|js|j |%t|} | j |r#|t |d} |j || zd|j ||jdzdj|S)a[Adjust ``new_string`` so its indentation matches ``file_region``. Used after a non-exact fuzzy match: the LLM may have sent old_string and new_string with a different indent than the file actually has (e.g. 2-space indent in tool args vs 4-space indent on disk). The fuzzy strategy successfully matched anyway, but writing ``new_string`` verbatim would corrupt the file's indentation. Approach: 1. For each non-blank line in ``new_string``, compute its indent *relative* to the shallowest non-blank line of ``old_string`` (the LLM's base indent). 2. Anchor that relative indent onto the file's actual base indent (the leading whitespace of the file_region's first non-blank line). 3. Re-emit each non-blank line as ``file_base + (line_indent - llm_base)``. Blank lines and lines less-indented than the LLM's base are anchored directly to the file's base indent. No-op cases (returns ``new_string`` unchanged): - file_region or old_string has no meaningful line - LLM base indent equals file base indent - new_string is empty NrV ) rYrTrWrXappend startswithr0lstriprK) rZrr old_first file_first old_indent file_indent out_linesrO line_indent remainders r_reindent_replacementrgs4 &z2I' 4JJ.$Y/J%j1K[ I  &?zz|   T " )$/  ! !* -S_-.I   [94 5   [4;;u+== >? 99Y rcd|vrd|vr|Sdjfd|D}|}d|vrd|vr|jdd}d|vrd|vr|jdd}|S)u Conditionally unescape ``\t``/``\r`` in new_string. LLMs frequently send the two-character sequences ``\t`` (backslash + t) and ``\r`` (backslash + r) inside JSON tool-call arguments where they meant a real tab or carriage-return byte. Writing the string verbatim corrupts tab-indented files with literal backslash-letter pairs. The unescape is only applied per-sequence when the *matched region of the file* actually contains the corresponding control character — that is, we only convert ``\t`` -> tab when the file region we're replacing contains a real tab byte. Files that legitimately contain the literal two-character string ``"\t"`` (e.g. a Python source line that defines ``sep = "\t"``) get a backslash+t in the matched region instead of a tab, so we leave new_string alone. ``\n`` is intentionally excluded: newlines serialize correctly through JSON and rewriting backslash-n would corrupt escape sequences in string literals far more often than it would help. \t\rr@c3.K|] \}}||ywrBrCrDs rrHz-_maybe_unescape_new_string..6rIrJrQ )rKr)rrr8rLouts ` rr2r2st0 J5 #:ggK7KKO C |/kk%& |/kk%& Jrczdjfd|D}t|}t|}||k7r|St|}i}t|ddD]\} } | |vs | || <t d||} | j } g} | D]\}}}}}|dk(r[|j |d}|}|t|kr$|||kr|dz }|t|kr |||kr| j|||i|dk(r| j||||d k(r|d k(s| j|||dj| S) uPreserve Unicode characters from the file in the replacement string. When strategy 7 (unicode_normalized) matched, the file has Unicode characters (em-dashes, smart quotes, ellipsis, non-breaking spaces) but old_string/new_string from the LLM are ASCII equivalents. Writing new_string verbatim would silently corrupt the file's Unicode — em-dashes become two hyphens, smart quotes become straight quotes. This function aligns the replacement with the file's actual Unicode by diffing old_string→new_string and applying only the actual edits to the file's original text, preserving Unicode for unchanged portions. r@c3.K|] \}}||ywrBrCrDs rrHz3_preserve_unicode_in_replacement..QsG'%,GrJNequalrr&rdeleteinsert) rKr_build_orig_to_norm_map enumerater get_opcodesgetr0r])rr8rrrZnorm_old norm_filefile_orig_to_normfile_norm_to_origorig_posnpsmopcodes result_partstagi1i2j1j2 orig_startorig_ends` rr3r3?s$''GwGGK"*-H";/I90 <(*!"3CR"89- " & &$, b !- x 4BnnG!L&3RR '>*..r15J!H3{++%h/"4A 3{++%h/"4    Jx @ A I     2b 1 2 H_  H_    2b 1 2!3$ 77<  rct|dd}|}|D]*\}}||||}t|||} n|} |d|| z||dz},|S)a Apply replacements at the given positions. Args: content: Original content matches: List of (start, end) positions to replace new_string: Replacement text old_string: When non-None, signals that the match came from a non-exact fuzzy strategy; ``new_string`` is re-indented to match the file's actual indentation before substitution. Returns: Content with replacements applied c |dSNrrCxs rz%_apply_replacements..s 1Q4rT)keyreverseN)sortedrg) rr8rrsorted_matchesresultrFrGrZadjusteds rr4r4sq$GFN F$: s  !!%,K,[*jQH!H(*VCD\9 : Mrpatterncg}d} |j||}|dk(r |S|j||t|zf|t|z}H)zStrategy 1: Exact string match.rrp)findr]r0)rrr8rFposs rr'r'saG E ll7E* "9  N S3w</01 c'l" rc |jdDcgc]}|j}}dj|}|jd}|Dcgc]}|j}}t|||||Scc}wcc}w)z Strategy 2: Match with line-by-line whitespace trimming. Strips leading/trailing whitespace from each line before matching. rV)rWrXrK_find_normalized_matches)rrrO pattern_linespattern_normalized content_linescontent_normalized_liness rr(r(s/6mmD.ABdTZZ\BMB=1MM$'M9FG GG $ 8# C Hs A;Bcbd}||}||}t||}|sgSt|||S)zC Strategy 3: Collapse multiple whitespace to single space. c0tjdd|S)Nz[ \t]+r )resubss r normalizez2_strategy_whitespace_normalized..normalizesvvia((r)r'_map_normalized_positions)rrrrcontent_normalizedmatches_in_normalizeds rr)r)sK)#7+"7+,,>@RS   %W.@BW XXrc |jd}|Dcgc]}|j}}|jdDcgc]}|j}}t||||dj|Scc}wcc}w)z Strategy 4: Ignore indentation differences entirely. Strips all leading whitespace from lines before matching. rV)rWr_rrK)rrrrOcontent_stripped_linesrs rr*r*sz MM$'M8EFdkkmFF/6}}T/BCtT[[]CMC # 6=) GCs A9A>c>d}||}||k(rgSt||S)zt Strategy 5: Convert escape sequences to actual characters. Handles \n -> newline, \t -> tab, etc. cf|jddjddjddS)Nz\nrVrirQrjrl)rrs runescapez-_strategy_escape_normalized..unescapes-yy%--eT:BB5$OOr)r')rrrpattern_unescapeds rr+r+s0 P!)G# 7$5 66rc n|jd}|sgS|dj|d<t|dkDr|dj|d<dj|}|jd}g}t|}t t||z dzD]}||||z}|j } | dj| d<t| dkDr| dj| d<dj| |k(sjt ||||zt|\} } |j| | f|S)z Strategy 6: Trim whitespace from first and last lines only. Useful when the pattern boundaries have whitespace differences. rVrr&rp)rWrXr0rKrangecopy_calculate_line_positionsr]) rrrmodified_patternrr8pattern_line_countrS block_lines check_lines start_posend_poss rr,r,sU MM$'M  %Q'--/M! =A)"-335 byy/MM$'MG]+ 3}%(::Q> ?1#Aa*<&<= "&&( $Q--/ A { a )"o335KO 99[ !%5 5!:q!&8"8#g," Iw NNIw/ 01 Nroriginalcg}d}|D]:}|j|tj|}|| t|ndz }<|j||S)uBuild a list mapping each original character index to its normalized index. Because UNICODE_MAP replacements may expand characters (e.g. em-dash → '--', ellipsis → '...'), the normalised string can be longer than the original. This map lets us convert positions in the normalised string back to the corresponding positions in the original string. Returns a list of length ``len(original) + 1``; entry ``i`` is the normalised index that character ``i`` maps to. rr&)r]rrwr0)rrnorm_posrrs rrtrt/s_FH9 ht$!1CIq89 MM( Mr orig_to_norm norm_matchesci}t|ddD]\}}||vs |||<g}t|dz }|D]D\}}||vr ||} | } | |kr|| |kr| dz } | |kr || |kr|j| | fF|S)zNConvert (start, end) positions in the normalised string to original positions.Nrpr&)rur0r]) rrnorm_to_orig_startr|rresultsorig_len norm_startnorm_endrrs r_map_positions_norm_to_origrDs *,' Sb(9:4( - -+3 x (4&(G< 1$H , / H / / ' 3 !l8&L -lLI  *73L &|\ BBrc 4t|}t|}|jd}t|dkrgS|dj}|dj}|jd}|jd}t|} g} t t|| z dzD]G} || j|k(s|| | zdz j|k(s7| j | Ig} t| } | dk(rdnd}| D]} | dkrd}nLdj || dz| | zdz }dj |dd}td ||j}||k\s]t|| | | zt|\}}| j ||f| S) z Strategy 8: Match by anchoring on first and last lines. Adjusted with permissive thresholds and unicode normalization. rVrrpr&?gffffff?g?N) rrWr0rXrr]rKrratior)rrrrr first_line last_linenorm_content_linesorig_content_linesrpotential_matchesrSr8candidate_count threshold similaritycontent_middlepattern_middlerrs rr.r.s &g.L%g.L &&t,M =A q!'')Jb!'')I&++D1 t,]+ 3)*-??!C D( q ! ' ' )Z 7 q#559 : @ @ Bi O  $ $Q '( G+,O (1,$I 1  "J"YY'9!A#a@R>RST>T'UVN!YY}Qr':;N(~~NTTVJ  "!:"Aq+='=s7|" Iw NNIw/ 01 Nrc |jd}|jd}|sgSg}t|}tt||z dzD]}||||z}d}t||D]G\} } t d| j | j j } | dk\sC|dz }I|t|dzk\sut||||zt|\} } |j| | f|S)z Strategy 9: Line-by-line similarity with 50% threshold. Finds blocks where at least 50% of lines have high similarity. rVr&rNg?r) rWr0rziprrXrrr])rrrrr8rrSrhigh_similarity_countp_linec_linesimrrs rr/r/s MM$'MMM$'M  G]+ 3}%(::Q> ?1#Aa*<&<= !"!-= +NFF!$  GMMOCd{%*% + !C $6$< .sIdCIMINc38K|]}t|dzywrrRrs rrHz,_calculate_line_positions..sED#d)a-Err&)summin)rrrrrrs rrrsNImKZ.HIIIEM)8,DEEIG.'*G g rrrc |jd}t|}g}tt||z dzD]O}dj||||z} | |k(s t ||||zt|\} } |j | | fQ|S)a Find matches in normalized content and map back to original positions. Args: content: Original content string content_lines: Original content split by lines content_normalized_lines: Normalized content lines pattern: Original pattern pattern_normalized: Normalized pattern Returns: List of (start, end) positions in the original content rVr&)rWr0rrKrr]) rrrrrpattern_norm_linesnum_pattern_linesr8rSblockrrs rrrs ,11$7./G 3/03DDqH I 1 21Q9J5JKL & &!:q!&7"7W" Iw NNIw/ 0 1 Nr normalizednormalized_matchesc |sgSg}d}d}|t|kr|t|kr||||k(r|j||dz }|dz }nt||dvr9||dk(r1|j||dz }|t|krA||dvr:|dz }n4||dvr|j||dz }n|j||dz }|t|kr|t|kr|t|kr.|jt||dz }|t|kr.i}i}t|D]\}} | |vr||| <||| <g} |D]\} |vr|} ntfdt|D} | dz |vr || dz dz} n| | z z} | t|kr;|| dz dk(r0| t|kr"|| dvr| dz } | t|kr|| dvr| j| t| t|f| S)z Map positions from normalized string back to original. This is a best-effort mapping that works for whitespace normalization. rr&r\r c34K|]\}}|k\s |ywrBrC)rErSnrs rrHz,_map_normalized_positions..PsV41aa:oQVs )r0r]rur)rrrrorig_idxnorm_idxrnorm_to_orig_endr|roriginal_matchesrrrrs @rrrs  LHH S] "x#j/'A H H!5 5    ) MH MH h 5 (Z-AS-H    ) MH#h-'HX,>e,KA  h 5 (    ) MH    ) MH' S] "x#j/'A, S] "C O,A  S] " ' 5.( - -+3 x (%-".  2L H + ++J7JV9\+BVVJ a<+ +'1 59H!X %:;H c*o %*X\*Bc*IS]*x/AU/JA S]*x/AU/J S3x=-I JK/L2 r context_lines max_resultscF|r|sy|j}|j|rsy|dj}|s6|Dcgc]#}|js|j%}}|sy|d}g}tD]L\} } | j} | std|| j } | dkDs:|j | | fN|sy|j d|d|} g}t}| D]\}}td||z tt|t|z|z}|f}||vrB|j|djfdt|z D}|j ||syd j|Scc}w) zFind lines in content most similar to old_string for "did you mean?" feedback. Returns a formatted string showing the closest matching lines with context, or empty string if no useful match is found. r@rNg333333?c|d SrrCrs rrz$find_closest_lines..s qter)rrVc3FK|]}|zdzdd|zyw)r&4dz| NrC)rEjrrFs rrHz%find_closest_lines..s9 qy1}R =#;"< = s!z --- ) splitlinesrXrurrr]sortsetmaxrr0addrKr)rrrr old_linesanchorl candidatesscoredrSrOstrippedrtopparts seen_ranges_line_idxrGrsnippetrrFs @@rfind_closest_linesr fs W%%'I&&(M Mq\   !F )2@Aaggiaggi@ @AF]+&4::< fh7==? 3; MM5!* % &  KKOK$ + C E%K  8Ax-/0#m$hY&?-&OPcl +  )) 3;'    W   >>%  QAs FFerror match_countc^|dk7ry|r|jdsyt||}|syd|zS)uReturn a '\n\nDid you mean...' snippet for plain no-match errors. Gated so the hint only fires for actual "old_string not found" failures. Ambiguous-match ("Found N matches"), escape-drift, and identical-strings errors all have ``match_count == 0`` but a "did you mean?" snippet would be misleading — those failed for unrelated reasons. Returns an empty string when there's nothing useful to append. rr@zCould not findz& Did you mean one of these sections? )r^r )r r rrhints rformat_no_match_hintrs@a (()9: j' 2D  6 ==r)FrB)r)&__doc__rtypingrrrrdifflibrrstrrboolintr<r1rTrYrgr2r3r4r'r(r)r*r+r,rtrr-r.r/rrrr rrCrrrs< 22#SScs  SS16jQCjQSjQcjQ)-jQ:?S(SV-YabeYf@f:gjQZ%#%U38_0E%%(%69%>Fsm%Pcc#> s> > > QT> B!3!(+!(,U38_(=!BE!HA! A!c3h0A!A!!$A! A!JFJtE#s(O/D$'5=c]NQHS34c3h3H$C#$uS#X:O(YSY3Y4cSVhCXY* C # $uSRUXBW  77s7tE#s(O?T7&''c'd5c?>S'Tcd3i*s)uS#X' %S/:C#CCU3PS8_@UC>5C5#5$uS#X:O5p S 3 4c3h;P NT#YC(+=@EJ3PS8_& c $s) 8 >C>%(>36>;>>r