CSC 107 - Programming for Science

1
CSC 107 -Programming for Science

• Lecture 17
• Data Files

2
Question of the Day

• Move one matchstick to produce a square

3
Todays Goal

• Todays lecture discusses handling data stored on
  disk
• Discuss importance and utility of reading
• C commands to open and close files, read contents
  of opened files
• Also discuss data file parsing strategies
• After todays lecture, you should be ready to
  write programs that reads processes data from a
  file

4
Data Files

• Computers are fast but dumb
• Good at performing repetitive or complex tasks
  that combine simple instructions
• Large memory size good for data sets too large
  for humans
• Most everything relies on data files
• Searching through 1TB of saved web pages
• Processing sensors or lab equipment inputs
• Evaluating integrity of detailed design data
• Check models against actual results
• Automate drawing given description of motion

5
Working with Data Files

• File access in C requires file pointer
• Abstraction created to simplify improve
• Will not worry about details of how they work
• File pointers are just another data type
• Variables declared as FILE
• Our first non-numeric data!
• Everything today involves input/output
• includeltstdio.hgt needed to use functions

6
File Pointer Variables

• Need to be careful using declaring
• FILE fin // GoodFILE bob, jim //
  Bad! jim lacks powerFILE ham, eggs // Yes,
  now we are cooking!
• Variables only describe files
• Assignment to variable does not affect files
• Variables values meaningful only to C
• bob ham // Both use same file nowint i
  bob // Gets warning, is value ?

7
Opening a File

• To use a file, must first open the file
• FILE fopen(FILENAME, ACCESS)
• Both arguments must be in quotes
• FILENAME can include path
• Path is not required
• File names may be case-sensitive
• ACCESS specifies how file will be used
• r ? file will only be read from
• Reading starts at beginning of file

8
Opening a File (2)

• fopen returns NULL when unsuccessful
• NULL is symbolic constant defined in stdio.h
• May mean file does not exist, cannot be read, or
  some other reason
• Using NULL file pointer will crash program
• Returns some other value on success
• Value not meaningful to humans, anyway

9
Reading From File

• First assign file pointer to the opened file
• int fscanf(FILE, ...) reads from file
• First argument is file pointer to read from
• Rest of arguments identical to scanf

10
Reading From File (2)

• Always been using fscanf
• Actually used to access keyboard
• FILE stdin declared assigned automatically
• stdin used to simplify reading from
  keyboarddouble a, bscanf(lf lf, a, b)
  is just shorthand fordouble a,
  bfscanf(stdin, lf lf, a, b)

11
How Much Data In the Window

• May not know how much data is in file
• Some uses require amount of data to vary
• File formats may change to add/remove data
• Upgrading sensors may produce more results
• Do not want to constantly rewrite code
• Have better things to do with your time
• Hard-coded values can be difficult to find, fix
• Reading past end of file can cause problems
• Results of these reads undefined
• May cause even more severe problems

12
Checking For End of File

• Check value returned by fscanf
• Returns number of data items it was able to read
• When less than number requested, probably atend
  of fileFILE fin fopen(scores.txt,r)whil
  e ((fscanf(fin, lf, score)) 1) sum
  score count 1printf(Average from d
  scores was .2lf\n, count, sum/count)

13
Checking for End of File (2)

• feof(FILE) also checks for end of file
• Returns true if parameter is at end of file
  otherwise returns false
• Can make easier to read codeFILE fin
  fopen(scores.txt,r)while (!feof(fin))
  fscanf(fin, lf, score) sum score
  count 1printf(Average from d scores was
  .2lf\n, count, sum/count)

14
Better Solution

• Unless carefully written, using end-of-file
  checks as limit can cause problems
• Could write record count at start of file
• Begin by reading in this count
• for loop then reads in that number of records
• Could also use sentinel value at end of file
• Keep reading until sentinel value is read
• Using sentinel value can cause same problems as
  with keyboard input

15
Lessons from Kindergarten

• Once finished using a file, close it
• Done using fclose(FILE) function
• Must re-open closed file pointer before using
• Normally closing file is not important
• Can be helpful if program crashes or file pointer
  reassigned, however
• Also, some systems allow only one program at a
  time to access a file

16
Your Turn

• Get into groups and complete daily activity

17
For Next Lecture

• Start week 7 weekly assignment
• Start programming assignment 2
• Will be posted to course web page tomorrow