Merge pull request #43 from thombashi/develop

Develop

README.rst

@@ -26,6 +26,7 @@ Feature
     - tuple
 - Create table(s) from:
     - CSV file/text
+    - HTML file/text
     - JSON file/text
     - Microsoft Excel :superscript:`TM` file
     - `Google Sheets <https://www.google.com/intl/en_us/sheets/about/>`_
@@ -197,14 +198,23 @@ Installation
 Dependencies
 ============
 
-Python 2.7 or 3.3+
+Python 2.7+ or 3.3+
 
--  `DataPropery <https://github.com/thombashi/DataProperty>`__ (Used to extract data types)
--  `jsonschema <https://github.com/Julian/jsonschema>`__
--  `pathvalidate <https://github.com/thombashi/pathvalidate>`__
--  `path.py <https://github.com/jaraco/path.py>`__
--  `six <https://pypi.python.org/pypi/six/>`__
--  `xlrd <https://github.com/python-excel/xlrd>`__
+Mandatory
+-----------------
+
+- `beautifulsoup4 <https://www.crummy.com/software/BeautifulSoup/>`__
+- `DataPropery <https://github.com/thombashi/DataProperty>`__ (Used to extract data types)
+- `jsonschema <https://github.com/Julian/jsonschema>`__
+- `pathvalidate <https://github.com/thombashi/pathvalidate>`__
+- `path.py <https://github.com/jaraco/path.py>`__
+- `six <https://pypi.python.org/pypi/six/>`__
+- `xlrd <https://github.com/python-excel/xlrd>`__
+
+Optional
+-----------------
+
+- `lxml <http://lxml.de/installation.html>`__ (Faster HTML convert if installed)
 
 
 Test dependencies

docs/conf.py

@@ -373,6 +373,16 @@
     If the open |attr_mode| is neither ``"w"`` nor ``"a"``.
 """
 
+rp_template = u"""
+.. |tnt_filename| replace:: %(filename)s
+.. |tnt_format_name| replace:: %(format_name)s
+.. |tnt_format_id| replace:: %(format_id)s
+.. |tnt_global_id| replace:: %(global_id)s
+.. |tnt_key| replace:: %(key)s
+.. |tnt_title| replace:: %(title)s
+.. |tnt_sheet| replace:: %(sheet)s
+"""
+
 rst_prolog = (
     rp_common +
     rp_builtin +
@@ -381,5 +391,6 @@
     rp_module +
     rp_raises +
     rp_attr +
-    rp_method
+    rp_method +
+    rp_template
 )

docs/pages/installation.rst

@@ -9,14 +9,23 @@ Installation
 Dependencies
 ============
 
-Python 2.7 or 3.3+
-
--  `DataPropery <https://github.com/thombashi/DataProperty>`__ (Used to extract data types)
--  `jsonschema <https://github.com/Julian/jsonschema>`__
--  `pathvalidate <https://github.com/thombashi/pathvalidate>`__
--  `path.py <https://github.com/jaraco/path.py>`__
--  `six <https://pypi.python.org/pypi/six/>`__
--  `xlrd <https://github.com/python-excel/xlrd>`__
+Python 2.7+ or 3.3+
+
+Mandatory
+-----------------
+
+- `beautifulsoup4 <https://www.crummy.com/software/BeautifulSoup/>`__
+- `DataPropery <https://github.com/thombashi/DataProperty>`__ (Used to extract data types)
+- `jsonschema <https://github.com/Julian/jsonschema>`__
+- `pathvalidate <https://github.com/thombashi/pathvalidate>`__
+- `path.py <https://github.com/jaraco/path.py>`__
+- `six <https://pypi.python.org/pypi/six/>`__
+- `xlrd <https://github.com/python-excel/xlrd>`__
+
+Optional
+-----------------
+
+- `lxml <http://lxml.de/installation.html>`__ (Faster HTML convert if installed)
 
 
 Test dependencies

docs/pages/introduction/feature.txt

@@ -9,6 +9,7 @@ Feature
     - tuple
 - Create table(s) from:
     - CSV file/text
+    - HTML file/text
     - JSON file/text
     - Microsoft Excel :superscript:`TM` file
     - `Google Sheets <https://www.google.com/intl/en_us/sheets/about/>`_

docs/pages/reference/loader.rst

@@ -43,6 +43,24 @@ CsvTableTextLoader class
     :undoc-members:
     :show-inheritance:
 
+
+HtmlTableFileLoader class
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: simplesqlite.loader.HtmlTableFileLoader
+    :inherited-members:
+    :undoc-members:
+    :show-inheritance:
+
+
+HtmlTableTextLoader class
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: simplesqlite.loader.HtmlTableTextLoader
+    :inherited-members:
+    :undoc-members:
+    :show-inheritance:
+
 
 JsonTableFileLoader class
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

requirements/requirements.txt

@@ -1,3 +1,4 @@
+beautifulsoup4
 DataProperty>=0.8.1
 jsonschema
 pathvalidate>=0.5.1

setup.py

@@ -23,7 +23,7 @@
 
 setuptools.setup(
     name="SimpleSQLite",
-    version="0.4.8",
+    version="0.5.0",
     url="https://github.com/thombashi/SimpleSQLite",
     bugtrack_url="https://github.com/thombashi/SimpleSQLite/issues",
 
@@ -32,7 +32,7 @@
     description=summary,
     include_package_data=True,
     install_requires=install_requires,
-    keywords=["SQLite", "CSV", "JSON", "Excel", "Google Sheets"],
+    keywords=["SQLite", "CSV", "Excel", "Google Sheets", "HTML", "JSON"],
     long_description=long_description,
     license="MIT License",
     packages=setuptools.find_packages(exclude=["test*"]),

simplesqlite/core.py

@@ -8,7 +8,6 @@
 import logging
 import os
 import sqlite3
-import sys
 
 import dataproperty
 import pathvalidate

simplesqlite/loader/__init__.py

@@ -4,7 +4,6 @@
 .. codeauthor:: Tsuyoshi Hombashi <gogogo.vm@gmail.com>
 """
 
-
 from __future__ import absolute_import
 
 from .error import ValidationError
@@ -13,6 +12,8 @@
 
 from .csv.core import CsvTableFileLoader
 from .csv.core import CsvTableTextLoader
+from .html.core import HtmlTableFileLoader
+from .html.core import HtmlTableTextLoader
 from .spreadsheet.excelloader import ExcelTableFileLoader
 from .spreadsheet.gsloader import GoogleSheetsTableLoader
 from .json.core import JsonTableFileLoader

simplesqlite/loader/acceptor.py

@@ -4,7 +4,6 @@
 .. codeauthor:: Tsuyoshi Hombashi <gogogo.vm@gmail.com>
 """
 
-
 from __future__ import absolute_import
 import abc
 

simplesqlite/loader/constant.py

@@ -0,0 +1,19 @@
+# encoding: utf-8
+
+"""
+.. codeauthor:: Tsuyoshi Hombashi <gogogo.vm@gmail.com>
+"""
+
+from __future__ import absolute_import
+
+
+class TableNameTemplate(object):
+    __FORMAT = "%({:s})s"
+    DEFAULT = __FORMAT.format("default")
+    FILENAME = __FORMAT.format("filename")
+    FORMAT_NAME = __FORMAT.format("format_name")
+    FORMAT_ID = __FORMAT.format("format_id")
+    GLOBAL_ID = __FORMAT.format("global_id")
+    KEY = __FORMAT.format("key")
+    TITLE = __FORMAT.format("title")
+    SHEET = __FORMAT.format("sheet")

simplesqlite/loader/csv/core.py

@@ -4,14 +4,12 @@
 .. codeauthor:: Tsuyoshi Hombashi <gogogo.vm@gmail.com>
 """
 
-
 from __future__ import absolute_import
 import csv
 
-import dataproperty
-import path
 import six
 
+from ..constant import TableNameTemplate as tnt
 from ..interface import TableLoader
 from .formatter import CsvTableFormatter
 
@@ -52,6 +50,10 @@ def __init__(self, source):
         self.quotechar = '"'
         self.encoding = "utf-8"
 
+    @property
+    def format_name(self):
+        return "csv"
+
     def _to_data_matrix(self):
         from dataproperty.type import FloatTypeChecker
 
@@ -76,26 +78,25 @@ class CsvTableFileLoader(CsvTableLoader):
 
     def __init__(self, file_path=None):
         super(CsvTableFileLoader, self).__init__(file_path)
-        self.table_name = "%(filename)s"
 
     def make_table_name(self):
         """
         |make_table_name|
 
-            ================  ===========================
-            format specifier  value after the replacement
-            ================  ===========================
-            ``%(filename)s``  filename
-            ================  ===========================
+            ===================  ==================================
+            format specifier     value after the replacement
+            ===================  ==================================
+            ``%(filename)s``     filename (without extention)
+            ``%(format_name)s``  ``csv``
+            ``%(format_id)s``    unique number in the same format
+            ``%(global_id)s``    unique number in all of the format
+            ===================  ==================================
 
         :return: Table name.
         :rtype: str
         """
 
-        self._validate()
-
-        return self.table_name.replace(
-            "%(filename)s", path.Path(self.source).namebase)
+        return self._make_file_table_name()
 
     def load(self):
         """
@@ -122,10 +123,17 @@ def load(self):
 
         return formatter.to_table_data()
 
+    def _get_default_table_name_template(self):
+        return tnt.FILENAME
+
 
 class CsvTableTextLoader(CsvTableLoader):
     """
     Concrete class of CSV text loader.
+
+    .. py:attribute:: table_name
+
+        Table name string. Defaults to ``%(format_name)s%(format_id)s``.
     """
 
     def __init__(self, text):
@@ -140,7 +148,9 @@ def load(self):
         :raises simplesqlite.loader.InvalidDataError:
             If the CSV data is invalid.
 
-        .. seealso:: :py:func:`csv.reader`
+        .. seealso::
+            :py:func:`csv.reader`
+            :py:meth:`~.CsvTableFileLoader.make_table_name`
         """
 
         self._validate()
@@ -152,3 +162,6 @@ def load(self):
         formatter.accept(self)
 
         return formatter.to_table_data()
+
+    def _get_default_table_name_template(self):
+        return "{:s}{:s}".format(tnt.FORMAT_NAME, tnt.FORMAT_ID)

simplesqlite/loader/csv/formatter.py

@@ -4,7 +4,6 @@
 .. codeauthor:: Tsuyoshi Hombashi <gogogo.vm@gmail.com>
 """
 
-
 from __future__ import absolute_import
 
 import dataproperty
@@ -38,6 +37,8 @@ def to_table_data(self):
             raise InvalidDataError(
                 "data row must be greater or equal than one")
 
+        self._loader.inc_table_count()
+
         yield TableData(
             self._loader.make_table_name(),
             header_list, data_matrix)

simplesqlite/loader/data.py

@@ -4,7 +4,6 @@
 .. codeauthor:: Tsuyoshi Hombashi <gogogo.vm@gmail.com>
 """
 
-
 from __future__ import absolute_import
 import collections
 

simplesqlite/loader/error.py

@@ -4,7 +4,6 @@
 .. codeauthor:: Tsuyoshi Hombashi <gogogo.vm@gmail.com>
 """
 
-
 from __future__ import absolute_import
 
 
@@ -14,7 +13,7 @@ class ValidationError(Exception):
     """
 
 
-class InvalidDataError(Exception):
+class InvalidDataError(ValueError):
     """
     Raised when data is invalid to load.
     """

simplesqlite/loader/formatter.py

@@ -4,10 +4,9 @@
 .. codeauthor:: Tsuyoshi Hombashi <gogogo.vm@gmail.com>
 """
 
-
 from __future__ import absolute_import
-
 import abc
+
 import six
 
 from .acceptor import LoaderAcceptor