o h1@sddlZddlmZddlmZddlTejjejjgZ ddZ ddZ d d Z d d Z d dZddZddZddZddZddZddZddZddZdd Zd!d"Zd#d$Zd%Zd&Zd'd(Zd)d*Zd+d,Zd-d.Zd/d0Z d1d2Z!dS)3N)Tree) tree_reader)*cCstdd}|s J|rJt|jdksJt|dks"Jtd|d}|r.J|s4Jt|jdks=Jt|dksEJtd|gd}|rRJ|rXJt|jdksaJt|d ksiJdS) Nfoo)labelrbar)rchildrenz (bar foo)bazz(baz (bar foo)))ris_leafis_preterminallenrstr)rrr rd/var/www/html/env_mimamsha/lib/python3.10/site-packages/stanza/tests/constituency/test_parse_tree.pytest_leaf_preterminal s        rcCsBd}t|}t|d}t|dksJt|dksJdS)N/((S (VP (VB Unban)) (NP (NNP Mox) (NNP Opal))))rz#[(VB Unban), (NNP Mox), (NNP Opal)])r read_treeslistyield_preterminalsr r)texttrees preterminalsrrrtest_yield_preterminalss  rcCs:d}t|}|ddksJ|ddksJdS)Nz5(foo) ((S (VP (VB Unban)) (NP (NNP Mox) (NNP Opal))))rr )rrdepthrrrrr test_depth&s rcC0d}t|}t|}gd}||ksJdS)z Test getting the unique labels from a tree Assumes tree_reader works, which should be fine since it is tested elsewhere z((SBARQ (WHNP (WP Who)) (SQ (VP (VBZ sits) (PP (IN in) (NP (DT this) (NN seat))))) (. ?))) ((SBARQ (WHNP (WP Who)) (SQ (VP (VBZ sits) (PP (IN in) (NP (DT this) (NN seat))))) (. ?))))NPPPROOTSBARQSQVPWHNPN)rrrget_unique_constituent_labels)rrlabelsexpectedrrrtest_unique_labels,s   r*cCr)z2 Test getting the unique tags from a tree Z((SBARQ (WHNP (WP Who)) (SQ (VP (VBZ sits) (PP (IN in) (NP (DT this) (NN seat))))) (. ?)))).DTINNNVBZWPN)rrrget_unique_tags)rrtagsr)rrrtest_unique_tags:   r4cCr)3 Test getting the unique words from a tree r+)?WhoinseatsitsthisN)rrrget_unique_wordsrrwordsr)rrrtest_unique_wordsGr5r@cC2d}t|}t|d}gd}||ksJdS)r6((SBARQ (WHNP (WP Who)) (SQ (VP (VBZ sits) (PP (IN in) (NP (DT this) (NN seat))))) (. ?))) ((SBARQ (NP (DT this) (NN seat)) (. ?)))g?)r8r9r;N)rrrget_rare_wordsr>rrrtest_rare_wordsS   rDcCrA)r6rBr)r7r:r<N)rrrget_common_wordsr>rrrtest_common_words_rErGcCsld}t|}dgt|ksJd}t|}dgt|ks"Jd}t|}ddgt|ks4JdS)Nzj( (SBARQ-FOO (WHNP-BAR (WP Who)) (SQ#ASDF (VP=1 (VBZ sits) (PP (IN in) (NP (DT this) (NN seat))))) (. ?)))r"a>( (SBARQ-FOO (WHNP-BAR (WP Who)) (SQ#ASDF (VP=1 (VBZ sits) (PP (IN in) (NP (DT this) (NN seat))))) (. ?)))( (SBARQ-FOO (WHNP-BAR (WP Who)) (SQ#ASDF (VP=1 (VBZ sits) (PP (IN in) (NP (DT this) (NN seat))))) (. ?)))( (SBARQ-FOO (WHNP-BAR (WP Who)) (SQ#ASDF (VP=1 (VBZ sits) (PP (IN in) (NP (DT this) (NN seat))))) (. ?)))z (FOO) (BAR)BARFOO)rrrget_root_labelsrrrrtest_root_labelsks   rKcCs^gd}gd}t||D]\}}t|}t|dksJ|d}|t|ks,Jq dS)N)z^((SBARQ (WHNP (WP Who)) (SQ (VP (VBZ sits) (PP (-NONE- in) (NP (DT this) (NN seat))))) (. ?)))z^((SBARQ (WHNP (-NONE- Who)) (SQ (VP (VBZ sits) (PP (IN in) (NP (DT this) (NN seat))))) (. ?)))zb((SBARQ (WHNP (WP Who)) (SQ (VP (VBZ sits) (PP (IN in) (NP (-NONE- this) (-NONE- seat))))) (. ?))))zW(ROOT (SBARQ (WHNP (WP Who)) (SQ (VP (VBZ sits) (PP (NP (DT this) (NN seat))))) (. ?)))zO(ROOT (SBARQ (SQ (VP (VBZ sits) (PP (IN in) (NP (DT this) (NN seat))))) (. ?)))F(ROOT (SBARQ (WHNP (WP Who)) (SQ (VP (VBZ sits) (PP (IN in)))) (. ?)))r r)ziprrr prune_noner)rr)tertreerrrtest_prune_nonezs  rRcCsHd}d}t|}dd|D}t|dksJ|t|dks"JdS)Nzf( (SBARQ-FOO (WHNP-BAR (WP Who)) (SQ#ASDF (VP=1 (VBZ sits) (PP (IN in) (NP (DT this) (- -))))) (. ?)))z[(ROOT (SBARQ (WHNP (WP Who)) (SQ (VP (VBZ sits) (PP (IN in) (NP (DT this) (- -))))) (. ?)))cSsg|]}|qSr)simplify_labels.0rOrrr sz(test_simplify_labels..r rrrr rrr)rrrrtest_simplify_labelss  rYcsTd}d}ddit|}fdd|D}t|dksJ|t|dks(JdS) NrLzD(ROOT (FOO (WHNP (WP Who)) (SQ (VP (VBZ sits) (PP (IN in)))) (. ?)))r#rIcg|]}|qSr)remap_constituent_labelsrT label_maprrrVz1test_remap_constituent_labels..r rrWrXrr\rtest_remap_constituent_labelss r_csXd}d}ddddt|}fdd|D}t|d ks J|t|d ks*JdS) NrLI(ROOT (SBARQ (WHNP (WP unban)) (SQ (VP (VBZ mox) (PP (IN opal)))) (. ?)))unbanmoxopal)r8r;r9crZr) remap_wordsrTword_maprrrVr^z0test_remap_constituent_words..r rrWrXrrertest_remap_constituent_wordss  rgcCsPd}d}gd}t|}t|dksJ|d}||}|t|ks&JdS)NrLr`)rarbrcr7r r)rrr replace_wordsr)rr) new_wordsrrQnew_treerrrtest_replace_wordss  rkcCsnd}t|}t|dgksJd}t|}t|gdks#Jd}t|}t|gdks5JdS)Nz((VP (VB Unban)))r"r%rL)r!r"r#r$r%r&zZ((VP (VB Unban))) (ROOT (SBARQ (WHNP (WP Who)) (SQ (VP (VBZ sits) (PP (IN in)))) (. ?))))rmrnrlrorp)rrrget_compound_constituentsrrrrtest_compound_constituentss   rrcCs`d}t|}t|dksJ|d}||ksJt|}|d}||us(J||ks.JdS)z Check one tree from the actual dataset for == when built with compound Open, this didn't work because of a silly bug a+(ROOT (S (NP (DT The) (NNP Arizona) (NNPS Corporations) (NNP Commission)) (VP (VBD authorized) (NP (NP (DT an) (ADJP (CD 11.5)) (NN %) (NN rate) (NN increase)) (PP (IN at) (NP (NNP Tucson) (NNP Electric) (NNP Power) (NNP Co.))) (, ,) (UCP (ADJP (ADJP (RB substantially) (JJR lower)) (SBAR (IN than) (S (VP (VBN recommended) (NP (JJ last) (NN month)) (PP (IN by) (NP (DT a) (NN commission) (NN hearing) (NN officer))))))) (CC and) (NP (NP (QP (RB barely) (PDT half)) (DT the) (NN rise)) (VP (VBN sought) (PP (IN by) (NP (DT the) (NN utility)))))))) (. .)))r rN)rrr )rrrQtrees2tree2rrr test_equalss    ruu (ROOT (IP (NP (NNP 证券法)) (VP (PP (IN 对) (NP (DNP (NP (NP (NNP 中国)) (NP (NN 证券) (NN 市场))) (DEC 的)) (NP (NN 运作)))) (, ，) (PP (PP (IN 从) (NP (NP (NN 股票)) (NP (VV 发行) (EC 、) (VV 交易)))) (, ，) (PP (VV 到) (NP (NP (NN 上市) (NN 公司) (NN 收购)) (EC 、) (NP (NN 证券) (NN 交易所)) (EC 、) (NP (NN 证券) (NN 公司)) (EC 、) (NP (NN 登记) (NN 结算) (NN 机构)) (EC 、) (NP (NN 交易) (NN 服务) (NN 机构)) (EC 、) (NP (NN 证券业) (NN 协会)) (EC 、) (NP (NN 证券) (NN 监督) (NN 管理) (NN 机构)) (CC 和) (NP (DNP (NP (CP (CP (IP (VP (VV 违法)))))) (DEC 的)) (NP (NN 法律) (NN 责任)))))) (ADVP (RB 都)) (VP (VV 作) (AS 了) (NP (ADJP (JJ 详细)) (NP (NN 规定))))) (. 。))) u (DNP (NP (CP (CP (IP (VP (ASDF (NP (NN 上市) (NN 公司) (NN 收购)) (EC 、) (NP (NN 证券) (NN 交易所)) (EC 、) (NP (NN 证券) (NN 公司)) (EC 、) (NP (NN 登记) (NN 结算) (NN 机构)) (EC 、) (NP (NN 交易) (NN 服务) (NN 机构)) (EC 、) (NP (NN 证券业) (NN 协会)) (EC 、) (NP (NN 证券) (NN 监督) (NN 管理) (NN 机构)))))))) (DEC 的)) cCs`tt}t|dks J|ddksJtt}t|dks$J|ddks.JdS)Nr r)rrCHINESE_LONG_LIST_TREEr count_unary_depth WEIRD_UNARY)rrrrtest_count_unaries"s  rzcCs<d}d}t|}t|dksJd|d|ksJdS)NrzZ(_ROOT (_S (_VP (_VB Unban )_VB )_VP (_NP (_NNP Mox )_NNP (_NNP Opal )_NNP )_NP )_S )_ROOTr z{:L}rrrr formatrXrrrtest_str_bracket_labels+s  r}cCs`d}t|}t|dksJ|dsJd}t|}t|dks&J|dr.JdS)Nrr rz+((S (VP (VB Unban)) (NP (Mox) (NNP Opal)))))rrr all_leaves_are_preterminalsrrrr test_all_leaves_are_preterminals3s  rcCs0d}d}t|d}d|}||ksJdS)z) Test the latex format for trees z;\Tree [.S [.NP Jennifer ] [.VP has [.NP nice antennae ] ] ]zK(ROOT (S (NP (NNP Jennifer)) (VP (VBZ has) (NP (JJ nice) (NNS antennae)))))rz{:T}N)rrr|)r)rQrrrr test_latex>s  rcCshd}t|}t|dksJd}d|d|ksJd}d|d|ks)J|dj|ks2Jd S) z> Pretty print a couple trees - newlines & indentation a(ROOT (S (VP (VB Unban)) (NP (NNP Mox) (NNP Opal)))) (ROOT (S (NP (DT The) (NNP Arizona) (NNPS Corporations) (NNP Commission)) (VP (VBD authorized) (NP (NP (DT an) (ADJP (CD 11.5)) (NN %) (NN rate) (NN increase)) (PP (IN at) (NP (NNP Tucson) (NNP Electric)))))))z?(ROOT (S (VP (VB Unban)) (NP (NNP Mox) (NNP Opal)))) z{:P}ra>(ROOT (S (NP (DT The) (NNP Arizona) (NNPS Corporations) (NNP Commission)) (VP (VBD authorized) (NP (NP (DT an) (ADJP (CD 11.5)) (NN %) (NN rate) (NN increase)) (PP (IN at) (NP (NNP Tucson) (NNP Electric))))))) r z {:O} {:O}Nr{)rrr)rrrtest_pretty_printHs rcCs>d}t|}t|dksJ|d}t|dksJdS)Nzx(ROOT (S (NP (PRP I)) (VP (VBP want) (S (VP (TO to) (VP (VB lick) (NP (NP (NNP Jennifer) (POS 's)) (NNS antennae))))))))r rzx(ROOT (S (VP (S (VP (VP (NP (NNS antennae) (NP (POS 's) (NNP Jennifer))) (VB lick)) (TO to))) (VBP want)) (NP (PRP I)))))rrr reverser)rr reversed_treerrr test_reversels   r)"pytest%stanza.models.constituency.parse_treerstanza.models.constituencyr stanza.testsmarkpipelinetravis pytestmarkrrrr*r4r@rDrGrKrRrYr_rgrkrrrurwryrzr}rrrrrrrrs:         8   $