Lecture 13 — Data from Files and Web Pages


  • Files on your computer
    • Opening and reading files (review)
    • Closing (new)
    • Writing files (new)
  • Accessing files across the web
  • Parsing basics
  • Parsing html

Our discussion is only loosely tied to Chapter 8 of the text.

Review — String operations often used in file parsing

Let’s review and go over some very common string operations that are particularly useful in parsing files.

  • Remove characters from the beginning, end or both sides of a string with lstrip, rstrip and strip:

    >>> x = "red! Let's go red! Go red! Go red!"
    >>> x.strip("red!")
    " Let's go red! Go red! Go "
    >>> x.lstrip("red!")
    " Let's go red! Go red! Go red!"
    >>> x.rstrip("red!")
    "red! Let's go red! Go red! Go "
    >>> "    Go red!      ".strip()
    'Go red!'

    Space is the character removed by default.

  • Split a string using a delimiter, and get a list of strings. Space is the default delimiter:

    >>> x = "Let's go red! Let's go red! Go red! Go red!"
    >>> x.split()
    ["Let's", 'go', 'red!', "Let's", 'go', 'red!', 'Go', 'red!', 'Go', 'red!']
    >>> x.split("!")
    ["Let's go red", " Let's go red", ' Go red', ' Go red', '']
    >>> x.split("red!")
    ["Let's go ", " Let's go ", ' Go ', ' Go ', '']

    It returns the strings before and after the delimiter string in a list.

  • Find the first location of a substring in a string, return -1 if not found. You can also optionally give a starting and end point to search from:

    >>> x
    "Let's go red! Let's go red! Go red! Go red!"
    >>> x.find('red')
    >>> x.find('Red')
    >>> x.find('red',10)
    >>> x.find('red',10,12)
    >>> 'red' in x
    >>> 'Red' in x

Opening and Reading Files

  • Given the name of a file as a string, we can open it to read:

    f = open('abc.txt')

    This is the same as

    f = open('abc.txt','r')
    • Variable f now “points” to the first line of file abc.txt.
    • The 'r' tells Python we will be reading from this file — this is the default.
  • We can read in data through three primary methods. First,

    line = f.readline()

    reads in the next line up to and including the end-of-line character, and “advances” f to point to the next line of file abc.txt.

  • By contrast,

    s = f.read()

    reads the entire remainder of the input file as a single string,

    • storing the one (big) string in s, and
    • advancing f to the end of the file!
  • When you are at the end of a file, f.read() and f.readline() will both return "" (empty string).

Reading the contents of a file

  • The most common way to read a file is as follows:

    f = open('abc.txt')
    for line in f:
        print line
  • This for loop will equivalent to the following:

    f = open('abc.txt')
    for each line in the file:
        line is assigned the string corresponding to the
           contents of the line, including the new line
  • You can combine the above steps into a single for loop:

    for line in open('abc.txt'):

Closing and Reopening Files

  • The code below closes and reopens a file

    f = open('abc.txt')
    # Insert whatever code is need to read from the file
    # and use its contents ...
    f = open('abc.txt')
  • f now points again to the beginning of the file.

  • This can be used to read the same file multiple times.

Writing to a File

  • In order to write to a file we must first open it and associate it with a file variable, e.g.

    f_out = open("outfile.txt","w")
  • The "w" signifies write mode which causes Python to completely delete the previous contents of outfile.txt (if the file previously existed).

  • It is also possible to use append mode:

    f_out = open("outfile.txt","a")

    which means that the contents of outfile.txt are kept and new output is added to the end of the file.

  • Write mode is much more common than append mode.

  • To actually write to a file, we use the write method:

    f_out.write("Hello world!")
    • Each call to write passes only a single string.
    • Unlike what happens when using print, spacing and newline characters are required explicitly
    • The string may be formatted
  • You must close the files you write! Otherwise, the changes you made will not be recorded!!


Part 1 Exercise

  1. Given the file census_data.txt:

    Location    2000    2011
    New York State  18,976,811  19,378,102
    New York City   8,008,686   8,175,133

    What are the value of variables line1, line2, line3, and line4 after the following code executes?

    f = open("census_data.txt")
    line1 = f.readline()
    line2 = f.read()
    line3 = f.readline()
    f = open("census_data.txt")
    line4 = f.readline()
  2. For the same data above, what does the following program produce?

    f = open('census_data.txt')
    s = f.read()
    line_list = s.split('\n')
    print len(line_list)
  3. Write code to print all the lines in the above file except for the header line (the first line).

  4. Given a file containing test scores, one per line, write Python code to write a second file with the scores output in decreasing order, one per line, with the index on each line. For example, if the input file contains:


    then the output file should contain:

    0: 98
    1: 83
    2: 75
    3: 66
    4: 21

    This can be done in 10 or fewer lines of Python code.

Opening Static Web Pages

  • We can use the urllib module to access web pages.

  • We did this with our very first “real” example:

    import urllib
    words_file = urllib.urlopen(words_url)
  • Once we have words_file we can use the read, readline, and close methods just like we did with “ordinary” files.

  • When the web page is dynamic, we usually need to work through a separate API (application program interface) to access the contents of the web site. Recall the Flickr example.


  • Before writing code to read a data file or to read the contents of a web page, we must know the format of the data in the file.
  • The work of reading a data file or a web page is referred to as parsing.
  • Files can be of a fixed well-known format
    • Python code
    • C++ code
    • HTML (HyperText Markup Language, used in all web pages)
    • JSON (Javascript Object Notation, a common data exchange format)
    • RDF (resource description framework)
  • Often there is a parser module for these formats that you can simply use instead of implementing them from scratch
  • For code, parsers check for syntax errors.

Short tour of data formats

  • Python code:

    • Each statement is on a separate line
    • Changes in indentation are used to indicate entry/exit to blocks of code, e.g. within def, for, if, while...
  • HTML: Basic structure is a mix of text with commands that are inside “tags” < ... >.


          <title>HTML example for CSCI-100</title>
          This is a page about <a href="http://python.org">Python</a>.
          It contains links and other information.
  • Despite the clean formatting of this example, html is in fact free-form, so that, for example, the following produces exactly the same web page:

    <html><head><title>HTML   example for CSCI-100</title>
    </head> <body> This is a page about <a
    href="http://python.org">Python</a>.  It contains   links
    and other   information. </body> </html>
  • JSON: used often with Python in many Web based APIs:

       "class_name": "CSCI 1100"
       , "lab_sections" : [
              { "name": "Section 01"
                 , "scheduled": "T 10AM-12PM"
                 , "location": "Sage 2704"
              , { "name": "Section 02"
                 , "scheduled": "T 12PM-2PM"
                 , "location": "Sage 2112"
              } ]

    Similar to HTML, spaces do not matter.

    • Simplejson is a simple module for converting between a string in JSON format and a Python variable:

      >>> import simplejson as sj
      >>> x = ' [ "a", [ "b", 3 ] ] '
      >>> sj.loads(x)
      ['a', ['b', 3]]

Parsing ad-hoc data formats - Regular tabular data

We will examine some simple formats that you have already seen in various homeworks.

  • Parsing files with fixed format in each line, delimited by a character

    Often used: comma (csv), tab or space

    • Example: lego list:

      2x1, 2 2x2, 3

    • Is there a header or not?

    Pseudo code:

    for each line of the file
        split using the separator
        read each column
  • Exercise: write a simple parser for the lego list that returns a list of the form:

    ['2x1', '2x1', '2x2', '2x2', '2x2']

Parsing ad-hoc data formats - Irregular tabular data

  • Parsing files with one line per row of information, different columns containing unknown amount of information seperated with a secondary delimiter

    • Example: Yelp from Lab 4:

      Meka’s Lounge|42.74|-73.69|407 River Street+Troy, NY 12180|http://www.yelp.com/biz/mekas-lounge-troy|Bars|5|2|4|4|3|4|5

    Information after column 5 are all reviews

    The address field is separated with a plus sign

    Pseudo code:

    for each line of the file
        split using the separator
        read column with secondary separator, split
        for each value in the column
            read value
  • Exercise: Return the address as a list of [street, city, state, zip] and the number of reviews for each line of the Yelp file.

Parsing ad-hoc data formats - Nontabular data that spans multiple lines

  • More complex file formats:

    • Blocks of data (of unknown length) separated by spaces making up a record:

      4590 - Friday, July 16, 2004
      Comments: Ken Jennings game 33.
      Frank McNeil: a facilities management specialist from Louisville, Kentucky
      Mary McCarthy: a homemaker from Las Vegas, Nevada
      Ken Jennings: a software engineer from Salt Lake City, Utah (whose 32-day cash winnings total $1,050,460)
      AMERICAN WRITERS | 8 days after publishing his first novel, "This Side of Paradise", he married Zelda Sayre | (F. Scott) Fitzgerald
      right: Ken
      Value: $200
      Number: 1
      AMERICAN WRITERS | His ability to imitate the family doctor earned him this playwright the nickname "Doc" | Neil Simon
      Wrong: Triple Stumper
      Frank: Who was Eugene O'Neill?
      Value: $400
      Number: 2

    Pseudo code:

    for each line in the file
        if the line in the same block as the previous
            add to the block of lines to process
            process the current block
            start a new block
  • Exercise: Write a simple piece of code to decide when a new block is reached. What is the initial value?


  • You should now be comfortable opening, reading, writing and closing files on local computers.
  • Once text (or HTML) files found on the web are opened, the same reading methods apply just as though the files were local. Binary files such as images require special modules.
  • Parsing a file requires understanding its format, which is, in a sense, the “language” in which it is written.
  • We will continue with file parsing exercises in Lab 7 and in HW 6.