%s r %test_parse_pdf_pages_is_nonempty_listz2TestParsePdf.test_parse_pdf_pages_is_nonempty_listGs:(9%&,,---6<< 1$$$r*cddlm}||}d|jvsJt|jdtsJ|jddkDsJy)u8parse_pdf 결과의 metadata에 page_count 키가 있다rr<rN)r$r9r"r-intr>s r &test_parse_pdf_metadata_has_page_countz3TestParsePdf.test_parse_pdf_metadata_has_page_countOsQ(9%v...&//,7===|,q000r*c|ddlm}tjt5|ddddy#1swYyxYw)u7잘못된 bytes를 넣었을 때 예외가 발생한다rr<s$this is not a valid pdf file contentN)r$r9pytestraises Exception)r'r9s r -test_parse_pdf_invalid_bytes_raises_exceptionz:TestParsePdf.test_parse_pdf_invalid_bytes_raises_exceptionXs/( ]]9 % ? = > ? ? ?s 2;r1) r2r3r4bytesr:r?rArDrIr5r*r r7r77sS/U/t/$u$$%u%%11$1?r*r7c4eZdZdeddfdZddZdeddfdZy)TestParseDocumentr rNcvddlm}m}||d}t||sJt |j dkDsJy)uYparse_document가 .pdf 파일명으로 포맷을 감지하여 ParseResult를 반환한다rrparse_documentz document.pdfN)r$rrOr-r%rr'r rrOr(s r +test_parse_document_detects_pdf_by_filenamez=TestParseDocument.test_parse_document_detects_pdf_by_filenameds8: >:&+...6;;!###r*cddlm}tjtt f5|dddddy#1swYyxYw)u:지원하지 않는 확장자는 예외를 발생시킨다r)rOs some byteszfile.xyzN)r$rOrFrG ValueErrorrH)r'rOs r 0test_parse_document_unsupported_extension_raiseszBTestParseDocument.test_parse_document_unsupported_extension_raisesls5- ]]J 2 3 6 =* 5 6 6 6s 9AcBddlm}m}||d}t||sJy)uB확장자 대소문자 구분 없이 포맷 감지가 동작한다rrNz document.PDFN)r$rrOr-rPs r .test_parse_document_case_insensitive_extensionz@TestParseDocument.test_parse_document_case_insensitive_extensionss!: >:&+...r*r1)r2r3r4rJrQrTrVr5r*r rLrLcs.$U$t$6//RV/r*rLc 8dddddddgdddddd dgd d d gd S)u>opendataloader-pdf JSON 출력을 흉내 내는 dict를 반환ztest.pdfr#Nheading)rrdrYu테스트 PDF 텍스트)typeid page number bounding boxz heading levelcontent paragraphr)rrrY2u페이지1 텍스트)rZr[r\r]r^)z file nameznumber of pagesauthortitlekidsr5r5r*r _make_mock_odl_outputrd~sN " 0!"4  $ /1    r* output_datacfd}|S)uPopendataloader_pdf.convert()의 side_effect: JSON 파일을 output_dir에 생성c<tj|jddt|tr|n|d}tj|j dz}tj||z j tjdy)NT)parentsexist_okrz.jsonF) ensure_ascii) pathlibPathmkdirr-r=stem write_textjsondumps) input_path output_dirkwargssrcfnameres r side_effectz._mock_convert_side_effect..side_effectsv Z &&td&C&z37jZ] S!&&0 j !E )55 JJ{ 7 r*r5)rerws` r _mock_convert_side_effectrxs  r*cteZdZdejdej ddfdZdejdej ddfdZdejdej ddfdZ dejdej ddfdZ dejdej ddfd Z dejdej ddfd Z y) TestParseCachetmp_path monkeypatchrNcddl}ddlm}|j|d|dz t}t d5}t ||_d}||}||} dddj jk(sJ|j| jk(sJ|j| jk(sJjdk(sJy#1swYlxYw) u`같은 bytes로 parse_pdf 2회 호출 시 같은 결과 반환, convert 호출은 1회만 발생rNr< CACHE_DIR parse_cacheopendataloader_pdf.converts%PDF-1.4 fake pdf contentr#) r$r9setattrrdrrxrwrr r! call_count) r'r{r|r$r9re mock_convert file_bytesresult1result2s r "test_cache_hit_returns_same_resultz1TestParseCache.test_cache_hit_returns_same_results (J X 5MN+- / 0 ,L'@'ML $5J +G +G  ,||w||+++}} ---~~///&&!+++ , ,s #CC cddl}ddlm}|j|d|dz t}t d5}t ||_|d|ddddjd k(sJy#1swYxYw) uG다른 bytes는 다른 캐시 — 각각 convert 호출이 발생한다rNr<r~rrs%PDF-1.4 content As%PDF-1.4 content Brr$r9rrdrrxrwr)r'r{r|r$r9rers r test_cache_miss_different_bytesz.TestParseCache.test_cache_miss_different_bytess~ (J X 5MN+- / 0 -L'@'ML $ + , + , - &&!+++  - -s !A11A:cddl}ddlm}|j|d|dz t}d}t d5}t ||_||||d dddjd k(sJy#1swYxYw) uJ`use_cache=False`로 호출 시 캐시를 무시하고 다시 파싱한다rNr<r~rs%PDF-1.4 use_cache false testrF) use_cacherr)r'r{r|r$r9rerrs r #test_use_cache_false_forces_reparsez2TestParseCache.test_use_cache_false_forces_reparses (J X 5MN+- 5 / 0 3L'@'ML $ j ! jE 2  3&&!+++ 3 3s #A55A>cddl}ddlm}m}|dz }|j|d|t }t d5}t ||_|d|dddd|jsJt|jd } t| d k\sJ|t|jd } t| dk(sJy#1swYwxYw) u8`clear_cache()` 후 캐시 디렉토리가 비어 있다rN) clear_cacher9rr~rs%PDF-1.4 clear cache test As%PDF-1.4 clear cache test B*.jsonr#) r$rr9rrdrrxrwexistsr.globr%) r'r{r|r$rr9 cache_dirrer json_files remainings r test_clear_cache_removes_allz+TestParseCache.test_clear_cache_removes_alls 5}, J Y?+- / 0 6L'@'ML $ 4 5 4 5 6 !!!)..23 :!###  12 9~""" 6 6s !CCcHddl}ddlm}|dz }|j|d|t}d}t d5}t ||_||ddd|jsJt|jd} t| d k(sJy#1swYFxYw) u1캐시 파일이 실제 디스크에 생성된다rNr<rr~s%PDF-1.4 disk cache testrrr#) r$r9rrdrrxrwrr.rr%) r'r{r|r$r9rrerrrs r test_cache_stores_to_diskz(TestParseCache.test_cache_stores_to_disks (}, J Y?+- 0 / 0 "L'@'ML $ j ! " !!!)..23 :!### " "s BB!c4ddl}ddlm}|j|d|dz t}d}t d5}t ||_||}dddt d5} t || _||} | jdddj jk(sJ|j| jk(sJ|j| jk(sJ|j| jk(sJy#1swYxYw#1swYxYw)uX캐시된 결과와 원본 결과가 동일하다 (text, pages, tables, metadata 비교)rNr<r~rs%PDF-1.4 result match testr) r$r9rrdrrxrwassert_not_calledrr r!r") r'r{r|r$r9rerroriginal mock_no_callcacheds r #test_cached_result_matches_originalz2TestParseCache.test_cached_result_matches_originals (J X 5MN+- 2 / 0 -L'@'ML $ ,H - / 0 -L'@'ML $z*F  * * ,  - }} +++~~---&--///  FOO333 - -  - -sD$)DD D) r2r3r4rkrlrF MonkeyPatchrrrrrrr5r*r rzrzs, ,393E3E, ,0, ,393E3E, ,&, ,393E3E, ,,# #393E3E# #6$ $393E3E$ $,4 4393E3E4 4r*rz)__doc__rprk unittest.mockrrrFr fixturerJr rr7rLr/rdrxrzr5r*r rs  * @ 511:&?&?X//6t6 4 "M4M4r*