Pyinstaller打包spacy库时失败

1.问题描述

在使用pyinstaller打包包含spacy库的文件时出错，点击可执行文件exe时，程序闪退，通过在cmd中输入“.\文件名.exe”可以看到具体错误的原因，全部都是提示类似下图这样的错误，都是提示某些模块缺失，无法导入某些名字等

2.问题分析

首先查看了spacy的依赖库，如下图所示，发现是spacy有很多依赖库，而提示的错误基本也都是这些依赖库的问题，所以最后发现问题原因就是在主程序中没有显示的导入这些依赖库，因此打包时没有将这些依赖库给打包起来，从而造成exe文件运行失败。

3.解决办法

（1）步骤1
将隐式导入的库通过显示的方式导入，将.spec文件中的hiddenimports修改为

hiddenimports=['srsly.msgpack.util', 'cymen', 'thinc', 'srsly', 'requests', 'plac', 'preshed', 'wasabi', 'cymem', 'murmurhash', 'blis','cymem.cymem', 'preshed.maps', 'thinc.linalg', 'thinc.neural._aligned_alloc', 'blis.py', 'spacy.strings', 'pkg_resources.py2_warn', 'spacy.morphology','thinc.extra.search','spacy._align','spacy.matcher.dependencymatcher','spacy.syntax.nonproj','spacy.language','spacy.cli.debug_data','spacy.tokens','spacy.matcher.matcher','spacy.cli.download','spacy.lang.en.tag_map','spacy.displacy.templates','spacy.lang.en.lemmatizer._adverbs_irreg','spacy.pipeline','spacy.cli.converters.conll_ner2json','spacy.about','spacy.cli.pretrain','spacy.lang.en.lemmatizer._adjectives_irreg','spacy._ml','spacy.gold','spacy.tokens.token','spacy.lang.en.lemmatizer','spacy.matcher.phrasematcher','spacy.lang.en.tokenizer_exceptions','spacy.cli.link','spacy.displacy.render','spacy.lang.en.lemmatizer._adverbs','spacy.lang.en.morph_rules','spacy.cli.profile','spacy.cli.train','spacy.pipeline.hooks','spacy.lang.en.lemmatizer._verbs_irreg','spacy.cli.info','spacy.glossary','spacy.attrs','spacy.lang.en.syntax_iterators','spacy.compat','spacy.lang.tokenizer_exceptions','spacy.tokens.underscore','spacy.lang.en','spacy.symbols','spacy.pipeline.entityruler','spacy.syntax.arc_eager','spacy.vectors','spacy.lang.en.lemmatizer._nouns_irreg','spacy.matcher._schemas','spacy.pipeline.pipes','spacy.lang','spacy.errors','spacy.syntax.transition_system','spacy.lang.char_classes','spacy.cli.init_model','spacy.tokenizer','spacy.cli','spacy.lang.en.lex_attrs','spacy.lang.norm_exceptions','spacy.cli.evaluate','spacy.pipeline.functions','spacy.lexeme','spacy.lang.en.lemmatizer.lookup','spacy.syntax._beam_utils','spacy.util','spacy.cli.convert','spacy.syntax._parser_model','spacy.lang.en.norm_exceptions','spacy','spacy.cli.validate','spacy.parts_of_speech','spacy.vocab','spacy.cli.converters.conllu2json','spacy.cli.converters.iob2json','spacy.morphology','spacy.tokens._retokenize','spacy.kb','spacy.lang.en.stop_words','spacy.syntax.stateclass','spacy.syntax','spacy.lang.en.lemmatizer._lemma_rules','spacy.lang.en.lemmatizer._verbs','spacy.lang.punctuation','spacy.cli.package','spacy.scorer','spacy.cli.converters.jsonl2json','spacy.lang.en.lemmatizer._nouns','spacy.lang.en.lemmatizer._adjectives','spacy.cli.converters','spacy.tokens.doc','spacy.syntax.ner','spacy.lang.tag_map','spacy.matcher','spacy.lemmatizer','spacy.strings','spacy.tokens.span','spacy.displacy','spacy.syntax.nn_parser','spacy.lang.lex_attrs']

（2）步骤2
创建并保存下述.py文件至 → python环境地址\Lib\site-packages\PyInstaller\hooks，此目的是将spacy库所需的模型数据一起打包

from PyInstaller.utils.hooks import collect_data_files
 
datas = collect_data_files("en_core_web_sm")

（3）步骤3
在需打包的文件中使用下述方法导入spacy库
import en_core_web_sm
nlp = en_core_web_sm.load()

修改完成后，使用pyinstaller -D 文件名.spec即可打包成功，并能够成功使用spacy