<Copyright 2023> Dumitrescu Toma-Ioan

Image editor:

## Description:

The program resembles a photo editor for the formats pgm and ppm, binary and
text files, performing commands such as load, select, save, crop, equalize,
histogram, apply filters. The memory is administered efficiently, so that
there are no memory leaks (verified with valgrind debbuger).
Launching the program: 1) make	2)./image_editor

Verifying images:
convert sample.jpg sample.ppm (binary file conversion for color image)
convert sample.jpg -compress None sample.pgm (text file for grayscale image)

## Implementation:

Command menu: the operations are indentified from the command text read as a
string. If there are more parameters, for that command, the parameters will be
obtained using fget function to catch the whole line and verify the format.

Image memorizing technique: for easy implementation and program adaptability,
the magic number that indicates the format for image and the source file is
retained as a local variable in main, being passed as parameter in all command
functions. The selection x1, y1, x2, y2 that has the same memorization logic.
To retain the image matrix, there are two matrices in the image structure,
maximum one of them being allocated at a timestamp. If there is no image,
the program handles to set the magic_number correctly to -1, determining easy
memory management. For colored images, the structure pixel retains a 3-uple
(r, g, b) for every pixel. A potential issue is max_val considered 255, but
the program can be easily changed to a more general max_val. Also, the
structure is not necessarily efficient, because of the padding, but the
current implementation reduces the number of parameters and does not affect
the overall functionality because of the non-complex inputs the program gets.

Error handling: For each function, the magic number, the parameters line,
the format of the parameters are the indicators of errors that are reported
via stdout messages. Error priorities are set by looking at expected output.

Load: a function that reads the image data from a file with the magic
number in the set P2, P3, P5, P6. Firstly, the file is read as text to get
the magic number, delete the current stored image if the command was valid
(so no memory leaks related to image matrix), and ignore comments. The
current position from the start is retained, pointing to the beggining of
the matrix. If the file is found to be binary, the reading will start from
the image matrix (the retained position), by opening the file with a new
handler and calling fseek. In other case, the same text file is used. Each
time a valid load is called, there is a function that allocates the matrix with
the specific dimensions and types. The comments are read using fgets, but if
the next line is with important data, then the data is extracted using sscanf.
The cases when comments are between max_val and the matrix are neglected.

Save function traverses the image data and prints it to a writable file F,
attempting to create F if it does not exist.

Histogram: Firstly, the frequency of each value in the image matrix is
calculated. The bins signify the frequency segments in the interval [0 - 255],
the frequency of a segment being the sum of all frequencies of the values
appearing on that segment. The histogram is displayed using stars applying the
histogram formula.

Crop: Treating each case (colored, grayscale), the portion [x1, x2) intersected
with [y1, y2) is cut from the image matrix, replacing the current image
matrix. To avoid memory leaks, the original matrix is assigned to a double
pointer and freed using that auxiliary pointer. After the replace and the free,
the image dimensions are updated. The idea is to go with two loops, one for
rows (i: y1 -> y2 - 1) and one for columns (j: x1 -> x2 - 1), allocate the
new matrix with the dimensions of that selection, traverse the original matrix
with those two iterators (i and j), then assign to the new matrix by
substracting the initial displacement from 0 of the specific iterator.

Select: Firstly, the whole selection case is verified, then verify if there is
a specific set of coordinates. The parameters are checked to be integers so
that atoi returns valid coordinates. An error here will be handled by the
program with the reset of coordinates to the previous values, ending the
function. Also, if there are more ore less numbers than needed, invalid command
message will be printed. After that, the set is checked to be in bounds.

Equalize: Regardless the selection, the equalize operation is performed on the
whole image. Since in the formula for a new pixel p is used only the sum of
the frequencies of values from 0 to p, then the frequency array will be a
partial sum array with fq[p] += fq[0] + fq[1] + ... + fq[p - 1], where the
fq[p] is initialized with the frequency of value p in the image. Since the
image is rectangular, the area is the product of width and height. Every
pixel is changed with a new pixel using an equalization formula.

Apply: all apply operation types are done using change_pixels() function that
effectively calculates the new pixels, by using the convolution of a 3x3
submatrix centered on a pixel with the filter kernel image, then using round
and clamp (0 for negative numbers, 255 for more than max_val values). The
pixels on the border of the image (0 and last lines and columns) are left
unchanged, the logic for that being in main apply function. Where no
coefficients are used, the filter coefficient will be considered 1.0.

## Bibliography:
https://ocw.cs.pub.ro/courses/programare