使用Python提取和读取PDF文件中的内容全方案与示例代码

# 常用
pip install pypdf pdfminer.six pdfplumber pymupdf

# 表格
pip install "camelot-py[base]"     # 1.0+ 默认用 PDFium，无需 Ghostscript（Linux 下更易装）:contentReference[oaicite:8]{index=8}
pip install tabula-py               # 需 Java 8+ 运行环境 :contentReference[oaicite:9]{index=9}

# OCR 路线1：一站式
# macOS 可: brew install ocrmypdf ；Linux/Win 请看文档 :contentReference[oaicite:10]{index=10}

# OCR 路线2：Python 组合
pip install pdf2image pytesseract   # 还需安装 Poppler 与 Tesseract 可执行文件 :contentReference[oaicite:11]{index=11}

# 深入与底层
pip install pypdfium2 pikepdf       # PDFium 绑定 & 元数据/结构 :contentReference[oaicite:12]{index=12}

from pypdf import PdfReader

reader = PdfReader("input.pdf")
text = "\n".join((page.extract_text() or "") for page in reader.pages)
print(text)

from pdfminer.high_level import extract_pages
from pdfminer.layout import LTTextContainer, LTChar, LAParams

for page_layout in extract_pages("input.pdf", laparams=LAParams()):
    for element in page_layout:
        if isinstance(element, LTTextContainer):
            print(element.get_text().strip())

import pdfplumber

with pdfplumber.open("input.pdf") as pdf:
    for i, page in enumerate(pdf.pages, 1):
        # 文本（可调 x/y 容差，提取更平滑）
        t = page.extract_text(x_tolerance=1, y_tolerance=3) or ""
        print(f"--- Page {i} ---\n{t}\n")

        # 表格（简单尝试）
        for table in page.extract_tables():
            for row in table:
                print(row)

import fitz  # PyMuPDF

doc = fitz.open("input.pdf")
for page in doc:
    # “text”=纯文本；“blocks”=文本块；“dict/json/html”=结构化/富文本输出
    print(page.get_text("blocks"))
    links = page.get_links()          # 链接
    annots = [a.info for a in page.annots() or []]  # 注释

import fitz
doc = fitz.open("input.pdf")
for page_index, page in enumerate(doc):
    for img in page.get_images(full=True):
        xref = img[0]
        pix = fitz.Pixmap(doc, xref)
        if pix.n > 4:   # CMYK等转RGB
            pix = fitz.Pixmap(fitz.csRGB, pix)
        pix.save(f"img_p{page_index}_{xref}.png")

import camelot

# 自动模式（按页号），尝试 lattice 或 stream
tables = camelot.read_pdf("tables.pdf", pages="1-3", flavor="lattice")
print(tables.n)                 # 抽到了多少张表
df = tables[0].df               # 直接拿 pandas.DataFrame
tables.export("out.csv", f="csv", compress=True)

import tabula
dfs = tabula.read_pdf("tables.pdf", pages="all", multiple_tables=True)

ocrmypdf -l chi_sim+eng --rotate-pages --deskew input_scan.pdf searchable.pdf

from pdf2image import convert_from_path
import pytesseract

pages = convert_from_path("scan.pdf", dpi=300)  # 依赖 Poppler
full_text = []
for img in pages:
    txt = pytesseract.image_to_string(img, lang="chi_sim+eng")
    full_text.append(txt)
print("\n".join(full_text))

from tika import parser
parsed = parser.from_file("input.pdf")
print(parsed["content"])     # 纯文本
print(parsed["metadata"])    # 元数据

import pypdfium2 as pdfium

pdf = pdfium.PdfDocument("input.pdf")
page = pdf.get_page(0)
textpage = page.get_textpage()
# 搜索关键字，返回范围迭代器
for match in textpage.search("发票", match_case=False):
    # 获取该命中范围的包围盒（可高亮/裁剪区域抽取）
    rect = textpage.get_rect(match)
    print(rect.left, rect.top, rect.right, rect.bottom)

from pypdf import PdfReader

reader = PdfReader("has_attachments.pdf")
for name, blobs in reader.attachments.items():
    for i, content in enumerate(blobs):
        with open(f"{name}-{i}", "wb") as f:
            f.write(content)

# pypdf：读表单域与值
from pypdf import PdfReader
reader = PdfReader("form.pdf")
fields = reader.get_fields()        # 或 reader.get_form_text_fields()
print(fields)

# pypdf 读取注释类型与位置
from pypdf import PdfReader
r = PdfReader("annotated.pdf")
for page in r.pages:
    if "/Annots" in page:
        for a in page["/Annots"]:
            obj = a.get_object()
            print(obj["/Subtype"], obj["/Rect"])

import pikepdf
pdf = pikepdf.open("input.pdf")
print(pdf.docinfo)            # 旧式 DocumentInfo（PDF 2.0 已废弃但仍常见）
meta = pdf.open_metadata()    # XMP 元数据
print(meta)

import pdfplumber

with pdfplumber.open("input.pdf") as pdf:
    page = pdf.pages[0]
    # bbox = (x0, top, x1, bottom) ，单位：PDF points
    region = page.within_bbox((72, 72, 540, 200))
    print(region.extract_text())

"""
功能：
1) 先用 pypdf 试文本；太少/失败 -> 判断可能是扫描件 -> 走 OCRmyPDF 或 pdf2image+pytesseract
2) 可选：Camelot 抽表、PyMuPDF 抽图片/链接/注释、pypdf 抽附件，pikepdf 取元数据
3) 输出 JSON：text/table/images/annotations/forms/attachments/metadata
"""
import json, os, subprocess, tempfile, shutil
from pypdf import PdfReader
import fitz

def is_text_pdf(path, min_chars=200):
    try:
        reader = PdfReader(path)
        s = "".join((p.extract_text() or "") for p in reader.pages[:5])
        return len(s.strip()) >= min_chars
    except Exception:
        return False

def ocr_if_needed(path):
    if is_text_pdf(path):
        return path  # 原样返回
    # 尝试用 OCRmyPDF（若未安装，可改为 pdf2image+pytesseract）
    out = os.path.join(tempfile.gettempdir(), f"ocr_{os.path.basename(path)}")
    try:
        subprocess.run(
            ["ocrmypdf", "--skip-text", "-l", "chi_sim+eng", path, out],
            check=True, capture_output=True
        )
        return out
    except Exception:
        return path  # 回退：继续用原文件（避免中断）

def extract_all(path):
    path = ocr_if_needed(path)
    result = {"text": "", "tables": [], "images": [], "links": [], "annots": [],
              "attachments": [], "metadata": {}, "forms": {}}

    # 1. 文本（pypdf）
    r = PdfReader(path)
    result["text"] = "\n".join((p.extract_text() or "") for p in r.pages)

    # 2. 附件与表单
    try:
        result["attachments"] = list(r.attachments.keys())
    except Exception:
        pass
    try:
        result["forms"] = r.get_fields() or {}
    except Exception:
        pass

    # 3. 元数据（pikepdf 可更全面，这里用 pypdf 的 DocInfo 兜底）
    try:
        result["metadata"] = dict(r.metadata or {})
    except Exception:
        pass

    # 4. 图片/链接/注释（PyMuPDF）
    doc = fitz.open(path)
    for i, page in enumerate(doc):
        # 图片
        for img in page.get_images(full=True):
            result["images"].append({"page": i+1, "xref": img[0], "width": img[2], "height": img[3]})
        # 链接
        for lk in page.get_links():
            result["links"].append({"page": i+1, **lk})
        # 注释
        for a in page.annots() or []:
            result["annots"].append({"page": i+1, **(a.info or {})})

    return result

if __name__ == "__main__":
    data = extract_all("example.pdf")
    print(json.dumps(data, ensure_ascii=False, indent=2))