运行 来自 Matlab 命令行的特定单元格部分？
Run particular cell section from command line in Matlab?
我在脚本中手动循环遍历 Matlab 中的各种单元格（我们称之为 foo.m）：
%%
%Code for cell 1
%%
%Code for cell 2
在 Matlab 的命令行中，我希望能够有选择地 运行 单元格 2 中的代码。The documentation 只有关于如何以交互方式执行此操作的说明（例如，将光标在适当的单元格中，然后等等等等）。我想要命令行的东西，所以我可以做一些像 foo.runCell(1) 到 运行 上面单元格 1 中的代码。
如果没有办法，我就把单元格拆分成单独的scripts/functions。这不太方便，因为我处于 'grind out prototype very quickly' 编码模式，所以现在希望所有内容都在一个文件中。
这是我的丑陋代码，可以满足您的要求。
尽管我尽了最大的努力，但我无法找到让它工作的方法 "properly"（即没有模拟按键）。我希望这将有助于其他人尝试找到正确的方法。无论如何，这里是：
runCodeSectionInFile.m:
function runCodeSectionInFile(scriptFullPath,sectionNum)
%// TIP: You can use "which('scriptName')" to obtain the full path of a
%// script on your MATLAB path.
%%% // Temporary Workaround
import java.awt.Robot; %// part of temporary workaround
import java.awt.event.*; %// part of temporary workaround
RoboKey = Robot; %// part of temporary workaround
RoboKey.setAutoDelay(10);%// part of temporary workaround
%% // Test if the needed components are available (optional)
if ~matlab.desktop.editor.isEditorAvailable || ...
~com.mathworks.mde.editor.codepad.Codepad.isCodepadEnabled
error('MATLAB editor is N\A');
end
%% // Open and\or switch to the script file
%// Test if script is opened:
if ~matlab.desktop.editor.isOpen(scriptFullPath)
scriptDoc = matlab.desktop.editor.openDocument(scriptFullPath);
else %// in case the script is open, get a handle to it, and save it:
scriptDoc = matlab.desktop.editor.findOpenDocument(scriptFullPath);
%// Save the script before running (optional):
scriptDoc.save;
end
scriptDoc.goToLine(0); %// Position the cursor at the beginning of the file
%// NOTE1: - uses zero based indexing!
jEd = com.mathworks.mlservices.MLEditorServices.getEditorApplication ...
.openEditorForExistingFile(java.io.File(scriptFullPath));
jEd.getTextComponent.grabFocus; drawnow; %// part of temp fix
%// NOTE2: most of the above can be replaced with:
%// EDITOROBJ = matlab.desktop.editor.openAndGoToLine(FILENAME,LINENUM);
%% // Get the Codepad and the LineManager handles:
jCm = com.mathworks.mde.editor.codepad.CodepadActionManager ...
.getCodepadActionManager(jEd);
jCp = jEd.getProperty('Codepad');
jLm = jCp.getLineManager(jEd.getTextComponent,jCm);
%% // Advance to the desired section
jAc = com.mathworks.mde.editor.codepad.CodepadAction.CODEPAD_NEXT_CELL;
%// 'next-cell' Action
for ind1=1:sectionNum-1 %// click "advance" several times
%// <somehowExecute(jAc) OR jCp.nextCell() >
RoboKey.keyPress(KeyEvent.VK_CONTROL); %// part of temporary workaround
RoboKey.keyPress(KeyEvent.VK_DOWN); %// part of temporary workaround
end
RoboKey.keyRelease(KeyEvent.VK_DOWN); %// part of temporary workaround
RoboKey.keyRelease(KeyEvent.VK_CONTROL); %// part of temporary workaround
%% Execute section - equivalent to clicking "Run Section" once
jAc = com.mathworks.mde.editor.codepad.CodepadAction.CODEPAD_EVALUATE_CELL;
%// 'eval-cell' Action
%// <somehowExecute(jAc); OR jCp.evalCurrentCell(true) >
RoboKey.keyPress(KeyEvent.VK_CONTROL); %// part of temporary workaround
RoboKey.keyPress(KeyEvent.VK_ENTER); %// part of temporary workaround
RoboKey.keyRelease(KeyEvent.VK_CONTROL); %// part of temporary workaround
%% // Close the file (optional)
jEd.close;
%% // Return focus to the command line:
com.mathworks.mde.cmdwin.CmdWin.getInstance.grabFocus;
testScript.m:
%% // This is code block one
disp('This is section one');
%% // This is code block two
disp('This is section two');
%% // This is code block three
disp('This is section three');
对于 运行 演示，只需将两个文件放在同一个文件夹中，确保它在您的路径（或 cd）上，然后执行 runCodeSectionInFile(which('testScript.m'),1)。结果如下：
>> runCodeSectionInFile(which('testScript.m'),1)
This is section one
>> runCodeSectionInFile(which('testScript.m'),2)
This is section two
>> runCodeSectionInFile(which('testScript.m'),3)
This is section three
如果你得到 NPE，你可能还需要将 RoboKey.setAutoDelay(10); 调整到更大的数字。
来源：
Some old question I posted and its answers.
P.S。
我在 2014b 上写了这篇文章。其他（旧）版本可能需要调整...
编辑 1：
通过找到 Codepad 对象及其 TaggedLineManager 的句柄，我设法解开了这个谜题的另一块。当前的问题是大多数 Codepad 方法在 EDT 上必须是 运行（类似于 "UI thread"，即源自与界面的交互）。
Dev-iL 使用 java 等提供了一个很好的答案...我将在此处提供一个不使用 java 或编辑器，但读取文件并在请求时评估语句。
为了使其工作，它有一个先决条件，即文件已保存（我很欣赏这不是 运行ning cells/codeBlocks 交互的先决条件） .
无论如何，我认为这是一个 "quirky" 问题，我想我会添加一个答案。
这是我的脚本 (cellScript.m)，其中包含代码块（请注意，我在“%%”之后为每个块指定了一个名称：
%Note: for this method your NOT allowed to put comments at the end of lines
%% cellA
disp ( 'running cell A' );
disp ( 'finished cell A' );
%% cellB
disp ( 'running cell B' );
disp ( 'finished cell B' );
%% cellC
disp ( 'running cell C' );
for ii=1:100
aa(ii) = randi(1);
end
% cells can have comments
disp ( 'past 1st coment' );
% cells comments can be indented
disp ( 'past indented comment' );
%{
block
comment
%}
disp ( 'past block comment' );
% cells can have comments
% multiple lines of comments
% not lined up
%
%and empty
disp ( 'past multiple lines of comments' );
disp ( 'finished cell C' );
%% cellD
disp ( 'running cell D' );
disp ( 'finished cell D' );
%% cellE
disp ( 'running cell E' );
disp ( 'finished cell E' );
我创建了一个 class 来完成请求的工作（我称之为 cellRunner.m ）
classdef cellRunner < handle
properties ( SetAccess = private )
fileName
fileInfo
cellInfo
cellNames = {};
end
methods
function obj = cellRunner ( file ) % constructor
if nargin == 0
obj.fileName = 'cellScript.m'; % default file for testing
else
obj.fileName = file; % store user file
end
obj.parseFile(); % read the file into memory
end
function obj = parseFile ( obj )
if ~isempty ( obj.fileInfo ) % on parsing check to see if its been parsed before
if isequal ( obj.fileInfo, dir ( obj.fileName ) ) % Check date stamp (has cell file been modified
% disp ( 'file not changed - reading skipped' ); % if not skip
% reading
return
end
end
obj.fileInfo = dir ( obj.fileName ); % store file info
fid = fopen ( obj.fileName ); % open file for reading
if fid ~= -1
index = 0; % this is the index of each cell
inCell = false; % has it found a cell to start reading
lines = cell(0);
while ( true )
line = fgetl ( fid ); % read the line in the file
if line == -1; break; end % check for the end of the file
sLine = strtrim ( line ); % trim any white space
if length ( sLine ) > 2 && strcmp ( sLine(1:2), '%%' ) % check to see if its the start of a cell
if index > 0 % Store the last cell data
obj.cellInfo{index} = lines; % in class to run when required
end
index = index + 1; % increment the index
obj.cellNames{index} = strtrim ( sLine(3:end) ); % save the name of the cell
lines = cell(0); % re-initialise the lines var
inCell = true; % the start of the cells have been found
elseif inCell % if reading a cell array
lines{end+1} = line; % add each line to the lines var
end
end
if index > 0 % make sure and save the last cell when finished reading
obj.cellInfo{index} = lines;
end
fclose ( fid );
else
error ( 'cellRunner:fileError', 'unable to read file' );
end
end
function obj = runCell ( obj, arg )
% obj.runCell ( 'cellName' );
% obj.runCell ( index );
obj.parseFile(); % check that the file hasn't been changed
if ischar ( arg ) % if user provided a char then search for it
index = strcmp ( arg, obj.cellNames ); % find the index
if ~any ( index ) % check it was found
error ( 'cellRunner:notFound', '%s not found', arg );
end
else
index = arg; % if index is an integer (not checked - assumed if not char)
if index < 1 || index > length ( obj.cellInfo ) % check integer is valid
error ( 'cellRunner:notFound', 'Index %d not found', arg );
end
end
commands = obj.cellInfo{index}{1}; % start to build the command to execute.
inBlock = false;
for ii=2:length(obj.cellInfo{index}) % loop around - ignoring any commented lines.
nextLine = strtrim ( obj.cellInfo{index}{ii} );
if inBlock
if length ( nextLine ) == 2 && strcmp ( nextLine, '%}' );
inBlock = false;
end
continue
end
if length ( nextLine ) == 2 && strcmp ( nextLine, '%{' );
inBlock = true;
continue
end
if length ( nextLine ) >= 1 && strcmp ( nextLine(1), '%' )
continue;
end
commands = sprintf ( '%s;%s', commands, obj.cellInfo{index}{ii} ); % build a parge string to eval
end
evalin('base',commands); % eval the expression in the base workspace.
end
end
end
然后使用代码如下：
obj.cellRunner();
% Individual cells can be run in two ways:
% By providing the name of the cell (the string after the %%)
obj.runCell ( 'cellC' );
% By providing the index
obj.runCell ( 3 );
注意 回想一下，必须保存文件才能工作。
样本运行：
whos
obj = cellRunner ( 'cellScript.m' );
obj.runCell ( 'cellC' );
running cell C
past 1st coment
past indented comment
past block comment
past multiple lines of comments
finished cell C
whos
Name Size Bytes Class Attributes
aa 1x100 800 double
ans 1x1 112 cellRunner
ii 1x1 8 double
obj 1x1 112 cellRunner
注1 - 为什么handle class? 我继承自handle class 因为我只想要一份已读取的文件数据- 有关何时使用 value/handle classes.
的出色概述，请参阅 answer 1 in this question
为了使用命令行运行文件中的特定部分，您可以使用
echodemo(filename, index of section)
我在脚本中手动循环遍历 Matlab 中的各种单元格（我们称之为 foo.m）：
%%
%Code for cell 1
%%
%Code for cell 2
在 Matlab 的命令行中，我希望能够有选择地 运行 单元格 2 中的代码。The documentation 只有关于如何以交互方式执行此操作的说明（例如，将光标在适当的单元格中，然后等等等等）。我想要命令行的东西，所以我可以做一些像 foo.runCell(1) 到 运行 上面单元格 1 中的代码。
如果没有办法，我就把单元格拆分成单独的scripts/functions。这不太方便，因为我处于 'grind out prototype very quickly' 编码模式，所以现在希望所有内容都在一个文件中。
这是我的丑陋代码，可以满足您的要求。
尽管我尽了最大的努力，但我无法找到让它工作的方法 "properly"（即没有模拟按键）。我希望这将有助于其他人尝试找到正确的方法。无论如何，这里是：
runCodeSectionInFile.m:
function runCodeSectionInFile(scriptFullPath,sectionNum)
%// TIP: You can use "which('scriptName')" to obtain the full path of a
%// script on your MATLAB path.
%%% // Temporary Workaround
import java.awt.Robot; %// part of temporary workaround
import java.awt.event.*; %// part of temporary workaround
RoboKey = Robot; %// part of temporary workaround
RoboKey.setAutoDelay(10);%// part of temporary workaround
%% // Test if the needed components are available (optional)
if ~matlab.desktop.editor.isEditorAvailable || ...
~com.mathworks.mde.editor.codepad.Codepad.isCodepadEnabled
error('MATLAB editor is N\A');
end
%% // Open and\or switch to the script file
%// Test if script is opened:
if ~matlab.desktop.editor.isOpen(scriptFullPath)
scriptDoc = matlab.desktop.editor.openDocument(scriptFullPath);
else %// in case the script is open, get a handle to it, and save it:
scriptDoc = matlab.desktop.editor.findOpenDocument(scriptFullPath);
%// Save the script before running (optional):
scriptDoc.save;
end
scriptDoc.goToLine(0); %// Position the cursor at the beginning of the file
%// NOTE1: - uses zero based indexing!
jEd = com.mathworks.mlservices.MLEditorServices.getEditorApplication ...
.openEditorForExistingFile(java.io.File(scriptFullPath));
jEd.getTextComponent.grabFocus; drawnow; %// part of temp fix
%// NOTE2: most of the above can be replaced with:
%// EDITOROBJ = matlab.desktop.editor.openAndGoToLine(FILENAME,LINENUM);
%% // Get the Codepad and the LineManager handles:
jCm = com.mathworks.mde.editor.codepad.CodepadActionManager ...
.getCodepadActionManager(jEd);
jCp = jEd.getProperty('Codepad');
jLm = jCp.getLineManager(jEd.getTextComponent,jCm);
%% // Advance to the desired section
jAc = com.mathworks.mde.editor.codepad.CodepadAction.CODEPAD_NEXT_CELL;
%// 'next-cell' Action
for ind1=1:sectionNum-1 %// click "advance" several times
%// <somehowExecute(jAc) OR jCp.nextCell() >
RoboKey.keyPress(KeyEvent.VK_CONTROL); %// part of temporary workaround
RoboKey.keyPress(KeyEvent.VK_DOWN); %// part of temporary workaround
end
RoboKey.keyRelease(KeyEvent.VK_DOWN); %// part of temporary workaround
RoboKey.keyRelease(KeyEvent.VK_CONTROL); %// part of temporary workaround
%% Execute section - equivalent to clicking "Run Section" once
jAc = com.mathworks.mde.editor.codepad.CodepadAction.CODEPAD_EVALUATE_CELL;
%// 'eval-cell' Action
%// <somehowExecute(jAc); OR jCp.evalCurrentCell(true) >
RoboKey.keyPress(KeyEvent.VK_CONTROL); %// part of temporary workaround
RoboKey.keyPress(KeyEvent.VK_ENTER); %// part of temporary workaround
RoboKey.keyRelease(KeyEvent.VK_CONTROL); %// part of temporary workaround
%% // Close the file (optional)
jEd.close;
%% // Return focus to the command line:
com.mathworks.mde.cmdwin.CmdWin.getInstance.grabFocus;
testScript.m:
%% // This is code block one
disp('This is section one');
%% // This is code block two
disp('This is section two');
%% // This is code block three
disp('This is section three');
对于 运行 演示，只需将两个文件放在同一个文件夹中，确保它在您的路径（或 cd）上，然后执行 runCodeSectionInFile(which('testScript.m'),1)。结果如下：
>> runCodeSectionInFile(which('testScript.m'),1)
This is section one
>> runCodeSectionInFile(which('testScript.m'),2)
This is section two
>> runCodeSectionInFile(which('testScript.m'),3)
This is section three
如果你得到 NPE，你可能还需要将 RoboKey.setAutoDelay(10); 调整到更大的数字。
来源：
Some old question I posted and its answers.
P.S。 我在 2014b 上写了这篇文章。其他（旧）版本可能需要调整...
编辑 1：
通过找到 Codepad 对象及其 TaggedLineManager 的句柄，我设法解开了这个谜题的另一块。当前的问题是大多数 Codepad 方法在 EDT 上必须是 运行（类似于 "UI thread"，即源自与界面的交互）。
Dev-iL 使用 java 等提供了一个很好的答案...我将在此处提供一个不使用 java 或编辑器，但读取文件并在请求时评估语句。
为了使其工作，它有一个先决条件，即文件已保存（我很欣赏这不是 运行ning cells/codeBlocks 交互的先决条件） .
无论如何，我认为这是一个 "quirky" 问题，我想我会添加一个答案。
这是我的脚本 (cellScript.m)，其中包含代码块（请注意，我在“%%”之后为每个块指定了一个名称：
%Note: for this method your NOT allowed to put comments at the end of lines
%% cellA
disp ( 'running cell A' );
disp ( 'finished cell A' );
%% cellB
disp ( 'running cell B' );
disp ( 'finished cell B' );
%% cellC
disp ( 'running cell C' );
for ii=1:100
aa(ii) = randi(1);
end
% cells can have comments
disp ( 'past 1st coment' );
% cells comments can be indented
disp ( 'past indented comment' );
%{
block
comment
%}
disp ( 'past block comment' );
% cells can have comments
% multiple lines of comments
% not lined up
%
%and empty
disp ( 'past multiple lines of comments' );
disp ( 'finished cell C' );
%% cellD
disp ( 'running cell D' );
disp ( 'finished cell D' );
%% cellE
disp ( 'running cell E' );
disp ( 'finished cell E' );
我创建了一个 class 来完成请求的工作（我称之为 cellRunner.m ）
classdef cellRunner < handle
properties ( SetAccess = private )
fileName
fileInfo
cellInfo
cellNames = {};
end
methods
function obj = cellRunner ( file ) % constructor
if nargin == 0
obj.fileName = 'cellScript.m'; % default file for testing
else
obj.fileName = file; % store user file
end
obj.parseFile(); % read the file into memory
end
function obj = parseFile ( obj )
if ~isempty ( obj.fileInfo ) % on parsing check to see if its been parsed before
if isequal ( obj.fileInfo, dir ( obj.fileName ) ) % Check date stamp (has cell file been modified
% disp ( 'file not changed - reading skipped' ); % if not skip
% reading
return
end
end
obj.fileInfo = dir ( obj.fileName ); % store file info
fid = fopen ( obj.fileName ); % open file for reading
if fid ~= -1
index = 0; % this is the index of each cell
inCell = false; % has it found a cell to start reading
lines = cell(0);
while ( true )
line = fgetl ( fid ); % read the line in the file
if line == -1; break; end % check for the end of the file
sLine = strtrim ( line ); % trim any white space
if length ( sLine ) > 2 && strcmp ( sLine(1:2), '%%' ) % check to see if its the start of a cell
if index > 0 % Store the last cell data
obj.cellInfo{index} = lines; % in class to run when required
end
index = index + 1; % increment the index
obj.cellNames{index} = strtrim ( sLine(3:end) ); % save the name of the cell
lines = cell(0); % re-initialise the lines var
inCell = true; % the start of the cells have been found
elseif inCell % if reading a cell array
lines{end+1} = line; % add each line to the lines var
end
end
if index > 0 % make sure and save the last cell when finished reading
obj.cellInfo{index} = lines;
end
fclose ( fid );
else
error ( 'cellRunner:fileError', 'unable to read file' );
end
end
function obj = runCell ( obj, arg )
% obj.runCell ( 'cellName' );
% obj.runCell ( index );
obj.parseFile(); % check that the file hasn't been changed
if ischar ( arg ) % if user provided a char then search for it
index = strcmp ( arg, obj.cellNames ); % find the index
if ~any ( index ) % check it was found
error ( 'cellRunner:notFound', '%s not found', arg );
end
else
index = arg; % if index is an integer (not checked - assumed if not char)
if index < 1 || index > length ( obj.cellInfo ) % check integer is valid
error ( 'cellRunner:notFound', 'Index %d not found', arg );
end
end
commands = obj.cellInfo{index}{1}; % start to build the command to execute.
inBlock = false;
for ii=2:length(obj.cellInfo{index}) % loop around - ignoring any commented lines.
nextLine = strtrim ( obj.cellInfo{index}{ii} );
if inBlock
if length ( nextLine ) == 2 && strcmp ( nextLine, '%}' );
inBlock = false;
end
continue
end
if length ( nextLine ) == 2 && strcmp ( nextLine, '%{' );
inBlock = true;
continue
end
if length ( nextLine ) >= 1 && strcmp ( nextLine(1), '%' )
continue;
end
commands = sprintf ( '%s;%s', commands, obj.cellInfo{index}{ii} ); % build a parge string to eval
end
evalin('base',commands); % eval the expression in the base workspace.
end
end
end
然后使用代码如下：
obj.cellRunner();
% Individual cells can be run in two ways:
% By providing the name of the cell (the string after the %%)
obj.runCell ( 'cellC' );
% By providing the index
obj.runCell ( 3 );
注意 回想一下，必须保存文件才能工作。
样本运行：
whos
obj = cellRunner ( 'cellScript.m' );
obj.runCell ( 'cellC' );
running cell C
past 1st coment
past indented comment
past block comment
past multiple lines of comments
finished cell C
whos
Name Size Bytes Class Attributes
aa 1x100 800 double
ans 1x1 112 cellRunner
ii 1x1 8 double
obj 1x1 112 cellRunner
注1 - 为什么handle class? 我继承自handle class 因为我只想要一份已读取的文件数据- 有关何时使用 value/handle classes.
的出色概述，请参阅 answer 1 in this question为了使用命令行运行文件中的特定部分，您可以使用
echodemo(filename, index of section)